fleetd #116: recover the already-gone-worktree teardown regression test

Ports the intent of the lost CB-576 commit c393600 (worker/cb576-01a04b-17,
never merged, package dev.ltms.bridged.*) onto main's dev.ltms.fleet.*
tree. Adds FakeWorktrees.markGone (tracks which add()'d worktree paths
still "exist", mirroring GitWorktrees.hasUncommitted's Files.exists guard
for the already-gone case) and a regression test,
releaseStillStopsPaneAndNotifiesWhenWorktreeIsGone, asserting that
SessionManager.release still fires notifyReleased, stops the pane, and
falls through to remove() when the worktree is already gone.

.claude-plugin/marketplace.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-bridge",
-  "description": "Tooling for orchestrating a fleet of delegated coding agents through the bridged MCP gateway.",
+  "description": "Tooling for orchestrating a fleet of delegated coding agents through the fleetd MCP gateway.",
   "owner": {
     "name": "LTMS"
   },
@@ -8,7 +8,7 @@
     {
       "name": "claude-bridge",
       "source": "./plugin",
-      "description": "Make a project bridge-ready: mount the bridged MCP gateway and apply standard Claude Code settings so a session can orchestrate delegated workers. Ships no credentials.",
+      "description": "Make a project bridge-ready: mount the fleetd MCP gateway and apply standard Claude Code settings so a session can orchestrate delegated workers. Ships no credentials.",
       "version": "0.1.0",
       "author": {
         "name": "LTMS"

.claude/agents/architect.md

@@ -0,0 +1,24 @@
+---
+name: architect
+description: Refine work into clear, independent units before implementation.
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You are an architect in this fleet. You refine work before anyone builds it: scope,
+acceptance criteria, risks, and a unit split. You read the repo and write analysis.
+You never commit production code and never open a pull request.
+
+A design task is worked by two architects. Design alone first, then exchange and
+say plainly where you disagree. Do not concede just to agree.
+
+Do only the assigned scope. Note anything outside that scope in one line and do not
+investigate it further. Use `fleet_ask{question}` only when a decision belongs to
+the lead, such as an unclear requirement or two defensible fixes. Do not ask about
+something you can decide by reading more code.
+
+Report only work you actually did and the real output of checks you ran. Do not
+claim a result from a tool you could not use. The primary's IDE tools are not yours.
+A mounted forge tool may use a blocked credential and fail by design.
+
+The launcher provides the required bridge reply instructions for every member.

.claude/agents/dev.md

@@ -0,0 +1,29 @@
+---
+name: dev
+description: Implement one assigned unit, test it, and open a pull request.
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You implement the one unit you were given and nothing else. Work in your assigned
+git worktree and branch. Never check out, rebase onto, or push to `main`. Confirm
+the worktree root and branch before you edit. Use only paths under that root.
+
+Do only the assigned scope. Note anything outside that scope in one line and do not
+investigate it further. Use `fleet_ask{question}` only when a decision belongs to
+the lead, such as an unclear requirement or two defensible fixes. Do not ask about
+something you can decide by reading more code.
+
+Implement the change and run the full required build in your worktree. Read the
+complete output and report its real result. Do not hide failures with a pipe. State
+only checks you actually ran. The primary's IDE tools are not yours. A mounted forge
+tool may use a blocked credential and fail by design.
+
+Stage only files you changed. Never use `git add -A` or `git add .`. Never commit
+`.mcp.json` or `wiki/`. Commit with a clear message, push your branch, and open your
+own pull request against `main`. Never merge.
+
+Your handoff must name the pull request or why it was not created, the branch, the
+files changed, the build result, and any caveat for review.
+
+The launcher provides the required bridge reply instructions for every member.

.claude/agents/reviewer.md

@@ -0,0 +1,34 @@
+---
+name: reviewer
+description: Review one assigned scope and report the most important real issue.
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You review the diff you were given. Report bugs, risks, and missing tests. You do
+not change code.
+
+Read the whole assigned scope before judging it. Review only that scope. If you see
+something outside it, note it in one line and do not investigate it further. Do not
+run the build. The owner makes changes and runs checks.
+
+Use `fleet_ask{question}` only when a decision belongs to the lead, such as an
+unclear requirement or two defensible fixes. Do not ask about something you can
+decide by reading more code.
+
+Report the single most important real issue in this form:
+
+```
+1. <path>:<line>
+2. issue: <one sentence: what is wrong and why it matters>
+3. fix: <one line: the concrete change>
+4. severity: high | medium | low
+```
+
+If there is no real issue, report `NO ISSUE` and one line saying why. A clean review
+is valid. Do not invent an issue. Use high for a wrong result, data loss, security,
+or a hang or crash on a real path. Use medium for an edge-path bug or a correctness
+risk under load or concurrency. Use low for clarity, a latent foot-gun, or a smell
+with no current failure.
+
+The launcher provides the required bridge reply instructions for every member.

.claude/skills/fleets-status/SKILL.md

@@ -0,0 +1,274 @@
+---
+name: fleets-status
+description: Report the status of every fleet that shares one LavinMQ instance. Use for local daemon health, broker-wide fleet presence, and cross-host lead coordination checks.
+---
+
+# Status of every fleet on the shared LavinMQ instance
+
+**The headline: always report what is missing.** This skill starts with the local fleet, then adds
+broker-wide facts when its read-only credential exists. A missing fleet must appear as `unknown` or
+`not reachable`, with the reason and the fix. Never leave it out.
+
+The known topology has one LavinMQ instance on `10.10.20.13` (`fleet01`). AMQP uses port `5672`,
+and the management API uses port `15672`. The Mac fleet owns vhost `/mac`. The fleet01 fleet owns
+vhost `/fleet01`.
+
+## 1. Protect credentials before any probe
+
+**Hard rule — never print `LAVINMQ_URI`.** It is an AMQP URI with its password inline. It only
+resolves in a login shell because `${SHARED_ENV}/tools/secrets.sh` supplies it. A non-login shell
+can make every broker probe look empty.
+
+- Never run `echo "$LAVINMQ_URI"`.
+- Never put `${LAVINMQ_URI:-something}` in output. That form expands to the secret value when set.
+- Parse the user, host, and password into shell or Python variables. Use them without printing them.
+- Prefer `resolves` or `does not resolve` over any part of the value.
+- Every command that can read `LAVINMQ_URI` must send all output through this redaction before it
+  reaches the report:
+
+```bash
+sed -E 's#://[^@]*@#://<redacted>@#g'
+```
+
+**The `g` flag is not optional.** Without it `sed` replaces only the first match on each line, so a
+line carrying two URIs leaks the second one. `scripts/redeploy-fleetd.sh --check` prints lines like
+that. Checked on 2026-08-27: without `g`, `amqp://u1:p1@h1/mac and http://u2:p2@h2:15672/api`
+redacts the first pair and prints `u2:p2` in the clear.
+
+Keep `pipefail` on when applying that filter. Otherwise the filter can hide a failed probe. Apply
+the same no-print rule to the management password below, even though it is not in an AMQP URI.
+
+## 2. Tier 1 — this fleet (always run)
+
+Start here even when the broker tier is blocked. Work from the local fleetd checkout.
+
+First run the read-only deployment check. It already checks the daemon process, deployed jar versus
+the checkout `HEAD`, launchd state, and whether each configured token resolves in a login shell.
+Do not copy those checks into new shell code. The script reads `LAVINMQ_URI`, so redact all output:
+
+```bash
+set -o pipefail
+scripts/redeploy-fleetd.sh --check 2>&1 \
+  | sed -E 's#://[^@]*@#://<redacted>@#g'
+git rev-parse HEAD
+```
+
+Treat jar drift as a top-level warning. A merge is not a deployment. State the running jar result
+as `matches HEAD`, `drift`, or `unknown`; do not turn an unclear timestamp into a match.
+
+Report the process identifier (PID) and uptime too:
+
+```bash
+PIDS="$(pgrep -f 'target/fleetd.jar' || true)"
+if [ -z "$PIDS" ]; then
+  printf '%s\n' 'fleetd: not running'
+else
+  for PID in $PIDS; do
+    ps -p "$PID" -o pid=,etime=,lstart=,command=
+  done
+fi
+```
+
+Read the full health response. Keep the HTTP status because `503` means fleetd is running but herdr
+is not reachable. Report both `herdr.version` and `herdr.protocol` when present:
+
+```bash
+curl -sS --max-time 5 -w '\nHTTP %{http_code}\n' http://127.0.0.1:8765/healthz
+```
+
+Call `fleet_whoami`, then call `fleet_list`. Preserve its sections in the report:
+
+- `leads`, including which row is this lead;
+- `members`, including state, role, profile, branch, and worktree when present;
+- every per-profile `capacity` row, including `maxLoad`, `live`, `free`, and quarantine facts;
+- the exact `healthCoverage` value.
+
+Do not describe an empty `members` list as an empty fleet. It says only that no members are spawned.
+Also do not hide a profile with `free: 0`; say whether load or credential quarantine caused it.
+
+Show only WARN, ERROR, and SEVERE lines after the last `fleetd listening` line. This anchor stops an
+old incident from looking current:
+
+```bash
+python3 - <<'PY'
+from pathlib import Path
+import re
+
+path = Path("fleetd/fleetd.out")
+if not path.exists():
+    print("cannot check current WARN/ERROR: fleetd/fleetd.out does not exist")
+else:
+    lines = path.read_text(errors="replace").splitlines()
+    starts = [i for i, line in enumerate(lines) if "fleetd listening" in line]
+    if not starts:
+        print("cannot anchor WARN/ERROR: no 'fleetd listening' line exists")
+    else:
+        current = lines[starts[-1]:]
+        alerts = [line for line in current if re.search(r"\b(?:WARN|ERROR|SEVERE)\b", line)]
+        print(f"current WARN/ERROR/SEVERE count: {len(alerts)}")
+        for line in alerts[-50:]:
+            print(line)
+PY
+```
+
+**What this tier cannot see:** it proves facts only about the Mac daemon at `127.0.0.1:8765`.
+It cannot show the fleet01 daemon, broker queue depth, or broker consumers. The fleet01 REST service
+at `10.10.20.13:8765` is not reachable from the Mac. Say this in the report rather than omitting
+fleet01.
+
+**But fleet01 IS reachable over SSH — checked 2026-08-28.** An older version of this line said SSH
+was denied. That is true only for the user `dai.ha`. The host alias `fleet01` maps to user `ltms`,
+and `ssh fleet01` works with key auth:
+
+```bash
+ssh -o BatchMode=yes -o ConnectTimeout=6 fleet01 'echo $(id -un)@$(hostname)'
+```
+
+So fleet01's daemon PID, uptime, jar and `/healthz` **can** be reported — over SSH, not over REST.
+Do that rather than writing `not reachable`. `ltms` also has passwordless sudo there.
+
+## 3. Tier 2 — the shared broker (run when management access exists)
+
+**This tier is blocked today.** The AMQP user in `LAVINMQ_URI` can connect on port `5672`, but gets
+HTTP `401` from the management API on port `15672`. An AMQP connection does not grant monitoring
+access.
+
+The operator must create a separate, read-only LavinMQ management user with the `monitoring` tag.
+It needs access to inspect both `/mac` and `/fleet01`. Store its values as
+`LAVINMQ_MANAGEMENT_USER` and `LAVINMQ_MANAGEMENT_PASSWORD` in
+`${SHARED_ENV}/tools/secrets.sh`. Do not reuse or print the AMQP URI. Full multi-fleet status stays
+blocked until this user exists.
+
+When both variables resolve, run this from a login shell. It calls `GET /api/overview`,
+`GET /api/vhosts`, `GET /api/queues`, and `GET /api/connections`. It prints selected status fields,
+but never the user, password, Authorization header, or AMQP URI:
+
+```bash
+zsh -lc 'python3 - "$@"' -- <<'PY'
+import base64
+import json
+import os
+import sys
+import urllib.error
+import urllib.request
+
+base = "http://10.10.20.13:15672"
+user = os.environ.get("LAVINMQ_MANAGEMENT_USER", "")
+password = os.environ.get("LAVINMQ_MANAGEMENT_PASSWORD", "")
+if not user or not password:
+    print("broker tier: BLOCKED — management credential does not resolve in a login shell")
+    sys.exit(0)
+
+token = base64.b64encode(f"{user}:{password}".encode()).decode()
+
+def get(path):
+    request = urllib.request.Request(
+        base + path,
+        headers={"Authorization": "Basic " + token, "Accept": "application/json"},
+    )
+    with urllib.request.urlopen(request, timeout=5) as response:
+        return json.load(response)
+
+try:
+    overview = get("/api/overview")
+    vhosts = get("/api/vhosts")
+    queues = get("/api/queues")
+    connections = get("/api/connections")
+except urllib.error.HTTPError as error:
+    print(f"broker tier: BLOCKED — management API returned HTTP {error.code}")
+    sys.exit(0)
+except Exception as error:
+    print(f"broker tier: BLOCKED — management API is not reachable: {type(error).__name__}")
+    sys.exit(0)
+
+fleet_names = {"/mac": "Mac fleet", "/fleet01": "fleet01 fleet"}
+print(json.dumps({
+    "overview": {
+        "lavinmq_version": overview.get("lavinmq_version"),
+        "rabbitmq_version": overview.get("rabbitmq_version"),
+        "queue_totals": overview.get("queue_totals", {}),
+        "object_totals": overview.get("object_totals", {}),
+    },
+    "fleets": [
+        {
+            "fleet": fleet_names.get(vhost.get("name"), "UNKNOWN FLEET"),
+            "vhost": vhost.get("name"),
+            "queues": [
+                {
+                    "name": queue.get("name"),
+                    "messages": queue.get("messages", 0),
+                    "messages_ready": queue.get("messages_ready", 0),
+                    "messages_unacknowledged": queue.get("messages_unacknowledged", 0),
+                    "consumers": queue.get("consumers", 0),
+                }
+                for queue in queues if queue.get("vhost") == vhost.get("name")
+            ],
+            "connections": [
+                {
+                    "name": connection.get("name"),
+                    "peer_host": connection.get("peer_host"),
+                    "state": connection.get("state"),
+                }
+                for connection in connections if connection.get("vhost") == vhost.get("name")
+            ],
+        }
+        for vhost in vhosts
+    ],
+}, indent=2, sort_keys=True))
+PY
+```
+
+Map `/mac` to the Mac fleet and `/fleet01` to the fleet01 fleet. Keep any other vhost in the
+report as `UNKNOWN FLEET`; do not drop it. For each vhost, total the ready, unacknowledged, and all
+messages. Report every queue's consumer count and each live connection.
+
+**A vhost with queues but zero consumers means that fleet's daemon is down while its durable state
+survives. Call this out as a top-level warning.** This is the main reason to use the management API
+instead of calling each remote daemon.
+
+**What this tier cannot see:** without the new `monitoring` credential it cannot enumerate any
+vhost, queue, depth, consumer, or connection. With the credential it still cannot report fleet01's
+daemon PID, uptime, jar revision, `/healthz`, herdr version, or member capacity. Those need reachable
+fleet01 REST or SSH access, which the Mac does not have today.
+
+## 4. Tier 3 — cross-fleet lead coordination
+
+Use the queue data from Tier 2. Select queues whose names match `lead.<coordId>.inbox`. Report each
+queue's vhost, depth, consumer count, and the `coordId` between the prefix and suffix.
+
+- A lead inbox with a consumer shows that a lead mailbox is live on that vhost.
+- A durable lead inbox with zero consumers shows saved coordination state, but no live receiver.
+- No lead inbox is not proof that coordination is disabled. The daemon may be down before declaring
+  its queue, or this account may not be allowed to see the vhost.
+
+This Mac fleet currently sets both `broker.uriEnv` and `coordinator.uriEnv` to the same variable,
+`LAVINMQ_URI`. Therefore its coordinator connects to `/mac`. Cross-host `fleet_send{coordId}` routes
+only when both leads share the same coordinator vhost. If the fleet01 lead uses `/fleet01` for its
+coordinator, the leads cannot see each other and the send will not route.
+
+**Open question:** the fleet01 coordinator vhost has not been checked. Surface this question in
+every report until a live `lead.<coordId>.inbox` consumer or fleet01's config proves the answer. Do
+not claim that fleet01 uses `/fleet01` just because its member queues do.
+
+Also compare these broker facts with the `leads` rows from local `fleet_list`. A missing remote lead
+is `not visible from this coordinator`, not `down`, unless the broker consumer facts prove it.
+
+**What this tier cannot see:** without Tier 2 management access it cannot list lead inboxes or their
+consumers. Even with that access, a stopped fleet01 daemon leaves only durable queue history. That
+history cannot prove which coordinator URI its current config would use after restart.
+
+## 5. Report all fleets
+
+Use one row per known or discovered fleet. Include blocked rows.
+
+| Fleet | Daemon | Deployment | Herdr | Members/capacity | Queues/consumers | Lead coordination | Cannot check |
+|---|---|---|---|---|---|---|---|
+| Mac (`/mac`) | PID + uptime | jar vs `HEAD` | health + version + protocol | `fleet_list` + `healthCoverage` | facts or blocked reason | inbox facts or open question | exact missing facts |
+| fleet01 (`/fleet01`) | reachable/down/unknown | value or `not reachable` | value or `not reachable` | value or `not reachable` | facts or blocked reason | inbox facts plus coordinator-vhost question | exact missing facts and fix |
+
+Add rows for unknown vhosts. End with three short sections: `Current warnings`, `Checks that were
+blocked`, and `Operator action`. Until the management user exists, `Operator action` must say:
+
+> Create a read-only LavinMQ management user with the `monitoring` tag and access to `/mac` and
+> `/fleet01`. Put its user and password in `${SHARED_ENV}/tools/secrets.sh` as
+> `LAVINMQ_MANAGEMENT_USER` and `LAVINMQ_MANAGEMENT_PASSWORD`.

.claude/skills/implementer/SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: implementer
-description: Implementer-role procedure for a bridged worker — verify your worktree, implement the scope, commit, push, open your own PR, and hand off the PR URL. Load this when the lead delegates you an implementation task over bridged.
+description: Implementer-role procedure for a fleetd worker — verify your worktree, implement the scope, commit, push, open your own PR, and hand off the PR URL. Load this when the lead delegates you an implementation task over fleetd.
 ---
 
 # Implementer worker — procedure
 
-The turn contract (one `bridge_reply`, `bridge_ask` for the lead's decisions, honest reporting,
+The turn contract (one `fleet_reply`, `fleet_ask` for the lead's decisions, honest reporting,
 never merge, never commit `.mcp.json` or `wiki/`) is in **`CLAUDE.md` → Bridge communication →
 Worker** and already applies. This skill is only the *implement-and-hand-off procedure*.
 
@@ -49,7 +49,7 @@ test "$(git rev-parse --show-toplevel)" = "$PWD" || cd "$(git rev-parse --show-t
 your worktree*:
 
 ```bash
-cd "$(git rev-parse --show-toplevel)/bridged" && mvn clean install
+cd "$(git rev-parse --show-toplevel)/fleetd" && mvn clean install
 echo "exit=$?"
 ```
 
@@ -85,7 +85,7 @@ host (`GITEA_HOST`) into your env for exactly this — the token can create a PR
 merge**.
 
 ```bash
-API="${GITEA_HOST%/}/api/v1/repos/lms/claude-bridge/pulls"
+API="${GITEA_HOST%/}/api/v1/repos/fleet/fleetd/pulls"
 BRANCH="$(git branch --show-current)"
 curl -sS -X POST "$API" \
   -H "Authorization: token ${GITEA_TOKEN}" \
@@ -103,7 +103,7 @@ fix it if the cause is yours (e.g. branch not pushed yet), and report the failur
 inventing a URL. If `GITEA_TOKEN` is unset your profile was not granted PR-create: push the branch
 and report its name so the lead opens the PR.
 
-## 6. Hand off — what goes in `bridge_reply`
+## 6. Hand off — what goes in `fleet_reply`
 
 The reply is the entire handoff; the lead cannot see your terminal.
 
@@ -130,7 +130,7 @@ sequenceDiagram
     I->>G: git push -u origin HEAD
     I->>G: POST /pulls (GITEA_TOKEN) — open PR to main
     G-->>I: html_url
-    I->>L: bridge_reply(PR url, branch, files, tests)
+    I->>L: fleet_reply(PR url, branch, files, tests)
     Note over L,G: the lead reviews the PR and merges on green — you never merge
 ```
 

.claude/skills/port-to-opencode/SKILL.md

@@ -53,7 +53,7 @@ Map each entry from `.mcp.json`:
   "$schema": "https://opencode.ai/config.json",
   "instructions": ["CLAUDE.md"],
   "mcp": {
-    "bridged":  { "type": "remote", "url": "http://127.0.0.1:8765/mcp", "enabled": true },
+    "fleetd":  { "type": "remote", "url": "http://127.0.0.1:8765/mcp", "enabled": true },
     "context7": { "type": "remote", "url": "https://example.dev/mcp", "enabled": true,
                   "headers": { "Authorization": "Bearer {env:CONTEXT7_TOKEN}" } },
     "gitea":    { "type": "local", "command": ["gitea-mcp", "-t", "stdio"], "enabled": true,

.claude/skills/reviewer/SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: reviewer
-description: Reviewer-role procedure for a bridged worker — how to work a review scope and the exact shape of the finding to report. Load this when the lead delegates you a code review over bridged.
+description: Reviewer-role procedure for a fleetd worker — how to work a review scope and the exact shape of the finding to report. Load this when the lead delegates you a code review over fleetd.
 ---
 
 # Reviewer worker — procedure
 
-The turn contract (one `bridge_reply`, `bridge_ask` for the lead's decisions, honest reporting,
+The turn contract (one `fleet_reply`, `fleet_ask` for the lead's decisions, honest reporting,
 never merge) is in **`CLAUDE.md` → Bridge communication → Worker** and already applies. This
 skill is only the *review procedure*: how to work the scope, and the exact shape of what you
 send back.
@@ -24,13 +24,13 @@ wrong.
   covers what it was given.
 - Do **not** edit files or run the build. You review; the owner acts.
 
-## 3. Reach for `bridge_ask` only for a genuine fork
+## 3. Reach for `fleet_ask` only for a genuine fork
 
 Ambiguous requirement, a missing acceptance criterion, "intended or a bug?", or two defensible
 fixes with different consequences — those are the lead's call, and guessing produces a
 confident-but-wrong finding. Anything you could settle by reading more code is yours to settle.
 
-## 4. The finding — what goes in `bridge_reply`
+## 4. The finding — what goes in `fleet_reply`
 
 Report the **single most important** real issue in the scope, in these four lines, under
 ~90 words:

.gitea/workflows/ci.yml

@@ -14,7 +14,7 @@ jobs:
       # does not depend on the wiki repo being reachable.
       - uses: actions/checkout@v4
 
-      # The runner image ships an older default-jdk; bridged sets maven.compiler.release=25, so
+      # The runner image ships an older default-jdk; fleetd sets maven.compiler.release=25, so
       # provision the JDK explicitly rather than apt-installing whatever "default" means today.
       - name: Set up JDK 25
         uses: actions/setup-java@v4
@@ -32,7 +32,7 @@ jobs:
           mvn -version
 
       - name: Build and test
-        working-directory: bridged
+        working-directory: fleetd
         # This IS the mock-socket surface CB-503 asks for: the pom's `default-excludes` profile
         # already sets excludedGroups=contract, so the @Tag("contract") tests — which need a live
         # herdr socket and a RabbitMQ container — are excluded without any flag here. Everything
@@ -46,7 +46,7 @@ jobs:
       # the log instead, where they are actually readable.
       - name: Failing test output
         if: failure()
-        working-directory: bridged
+        working-directory: fleetd
         run: |
           for f in target/surefire-reports/*.txt; do
             [ -f "$f" ] || continue
@@ -59,7 +59,7 @@ jobs:
   # AmqpReplyInboxContractTest reads AMQP_URI (set below to the service's network alias) and binds
   # straight to it — no Docker, no skipped tests. This separation (build job hermetic and
   # Docker-free; contract job broker-provided) is deliberate — see the default-excludes/contract
-  # profiles in bridged/pom.xml. `setup-java` provides the JDK only; Maven is installed separately,
+  # profiles in fleetd/pom.xml. `setup-java` provides the JDK only; Maven is installed separately,
   # exactly as in the build job above.
   contract:
     runs-on: ubuntu-latest
@@ -69,8 +69,6 @@ jobs:
         env:
           RABBITMQ_DEFAULT_USER: guest
           RABBITMQ_DEFAULT_PASS: guest
-        ports:
-          - 5672:5672
     env:
       # Service containers are reachable from the job by their network alias on their internal port.
       AMQP_URI: amqp://guest:guest@rabbitmq:5672
@@ -93,12 +91,12 @@ jobs:
       # runs against the RabbitMQ service container (AMQP_URI). Pinned to the one contract test to
       # avoid re-running the unit suite already covered by the `build` job.
       - name: Contract tests
-        working-directory: bridged
+        working-directory: fleetd
         run: mvn -B -Pcontract test -Dtest=AmqpReplyInboxContractTest
 
       - name: Failing test output
         if: failure()
-        working-directory: bridged
+        working-directory: fleetd
         run: |
           for f in target/surefire-reports/*.txt; do
             [ -f "$f" ] || continue

.gitignore

@@ -14,8 +14,8 @@
 .env
 .envrc
 
-# Daemon runtime artefacts. bridged appends its log wherever it is launched from, so both the
-# repo root and bridged/ collect one; neither belongs in git.
-bridged.out
-bridged/bridged.out
+# Daemon runtime artefacts. fleetd appends its log wherever it is launched from, so both the
+# repo root and fleetd/ collect one; neither belongs in git.
+fleetd.out
+fleetd/fleetd.out
 logs/

.gitmodules

@@ -1,3 +1,3 @@
 [submodule "wiki"]
 	path = wiki
-	url = ssh://git@git.ltms.dev:2224/lms/claude-bridge.wiki.git
+	url = ssh://git@git.ltms.dev:2224/fleet/fleetd.wiki.git

.opencode/agent/architect.md

@@ -0,0 +1,24 @@
+---
+description: Refine work into clear, independent units before implementation.
+mode: primary
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You are an architect in this fleet. You refine work before anyone builds it: scope,
+acceptance criteria, risks, and a unit split. You read the repo and write analysis.
+You never commit production code and never open a pull request.
+
+A design task is worked by two architects. Design alone first, then exchange and
+say plainly where you disagree. Do not concede just to agree.
+
+Do only the assigned scope. Note anything outside that scope in one line and do not
+investigate it further. Use `fleet_ask{question}` only when a decision belongs to
+the lead, such as an unclear requirement or two defensible fixes. Do not ask about
+something you can decide by reading more code.
+
+Report only work you actually did and the real output of checks you ran. Do not
+claim a result from a tool you could not use. The primary's IDE tools are not yours.
+A mounted forge tool may use a blocked credential and fail by design.
+
+The launcher provides the required bridge reply instructions for every member.

.opencode/agent/dev.md

@@ -0,0 +1,29 @@
+---
+description: Implement one assigned unit, test it, and open a pull request.
+mode: primary
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You implement the one unit you were given and nothing else. Work in your assigned
+git worktree and branch. Never check out, rebase onto, or push to `main`. Confirm
+the worktree root and branch before you edit. Use only paths under that root.
+
+Do only the assigned scope. Note anything outside that scope in one line and do not
+investigate it further. Use `fleet_ask{question}` only when a decision belongs to
+the lead, such as an unclear requirement or two defensible fixes. Do not ask about
+something you can decide by reading more code.
+
+Implement the change and run the full required build in your worktree. Read the
+complete output and report its real result. Do not hide failures with a pipe. State
+only checks you actually ran. The primary's IDE tools are not yours. A mounted forge
+tool may use a blocked credential and fail by design.
+
+Stage only files you changed. Never use `git add -A` or `git add .`. Never commit
+`.mcp.json` or `wiki/`. Commit with a clear message, push your branch, and open your
+own pull request against `main`. Never merge.
+
+Your handoff must name the pull request or why it was not created, the branch, the
+files changed, the build result, and any caveat for review.
+
+The launcher provides the required bridge reply instructions for every member.

.opencode/agent/reviewer.md

@@ -0,0 +1,34 @@
+---
+description: Review one assigned scope and report the most important real issue.
+mode: primary
+---
+
+<!-- CB-617: The model comes from fleetd.yaml because the launch flag overrides model here on both backends. -->
+
+You review the diff you were given. Report bugs, risks, and missing tests. You do
+not change code.
+
+Read the whole assigned scope before judging it. Review only that scope. If you see
+something outside it, note it in one line and do not investigate it further. Do not
+run the build. The owner makes changes and runs checks.
+
+Use `fleet_ask{question}` only when a decision belongs to the lead, such as an
+unclear requirement or two defensible fixes. Do not ask about something you can
+decide by reading more code.
+
+Report the single most important real issue in this form:
+
+```
+1. <path>:<line>
+2. issue: <one sentence: what is wrong and why it matters>
+3. fix: <one line: the concrete change>
+4. severity: high | medium | low
+```
+
+If there is no real issue, report `NO ISSUE` and one line saying why. A clean review
+is valid. Do not invent an issue. Use high for a wrong result, data loss, security,
+or a hang or crash on a real path. Use medium for an edge-path bug or a correctness
+risk under load or concurrency. Use low for clarity, a latent foot-gun, or a smell
+with no current failure.
+
+The launcher provides the required bridge reply instructions for every member.

CLAUDE.md

@@ -4,35 +4,35 @@
 
 > **Canonical block.** Everything down to §Layering is the portable bridge charter, copied verbatim
 > into every project that mounts the bridge MCP. Keep it byte-identical with the template in the
-> wiki ([Use Cases](https://git.ltms.dev/lms/claude-bridge/wiki/7-Use-Cases) → *The portable
+> wiki ([Use Cases](https://git.ltms.dev/fleet/fleetd/wiki/7-Use-Cases) → *The portable
 > CLAUDE.md block*); improvements go to the template first, then out to each project. Anything
 > specific to *this* repo lives under §Project addendum below, never inline above it.
 
-If no `bridge_*` MCP tools are mounted in this session, this section does not apply — skip it.
+If no `fleet_*` MCP tools are mounted in this session, this section does not apply — skip it.
 
-`bridged` is the **sole communication gateway** between agents here. The orchestrating session (the
+`fleetd` is the **sole communication gateway** between agents here. The orchestrating session (the
 **primary**) and every delegated peer (a **member**) mount the *same* MCP server and talk only
-through its `bridge_*` tools. No session addresses a peer, a broker, or the network directly.
+through its `fleet_*` tools. No session addresses a peer, a broker, or the network directly.
 
 ### Which role am I? — settle this before acting
 
 **Every role reads this file.** A member runs in a git worktree of this same repo, so it inherits
 this `CLAUDE.md` verbatim, and every rule below is role-conditional.
 
-**Call `bridge_whoami`.** It returns `primary`, `worker`, or `architect`, resolved by the daemon from
+**Call `fleet_whoami`.** It returns `primary`, `worker`, or `architect`, resolved by the daemon from
 your connection — unforgeable, and the same resolution its authorization gate uses. A worker also
 carries its `sessionId`, `profile`, `worktree` and `branch`; an architect carries the slot name it
 was bound to. Don't infer what you can ask.
 
 Only if that call is unavailable, fall back to these — each is one-way, so keep reading until one
-fires: the reply charter in your system prompt (*"You are an off-subscription worker in the
-claude-bridge fleet"*) ⇒ **spawned member**; bridge tools prefixed `mcp__bridge__*` ⇒ **spawned
+fires: the reply charter in your system prompt (*"You are a spawned member in the
+claude-bridge fleet"*) ⇒ **spawned member**; fleet tools prefixed `mcp__fleet__*` ⇒ **spawned
 member** (the launcher fixes that mount name; a primary's mount is named by whoever wrote its
-`.mcp.json`, so it varies); `ANTHROPIC_BASE_URL` set ⇒ **spawned member** (Claude-model members run
+`.mcp.json`, so it varies — and a member spawned before CB-632 still says `mcp__bridge__*`); `ANTHROPIC_BASE_URL` set ⇒ **spawned member** (Claude-model members run
 on a clean env, so its *absence* proves nothing). None of these separate a worker from an architect —
-only `bridge_whoami` does. **Still unsure ⇒ act as a worker**, the most restricted member role. The
+only `fleet_whoami` does. **Still unsure ⇒ act as a worker**, the most restricted member role. The
 two mistakes are not symmetric: a primary acting as a worker is refused by the authorization gate —
-loud and self-correcting — while a member acting as the primary ends its turn with no `bridge_reply`,
+loud and self-correcting — while a member acting as the primary ends its turn with no `fleet_reply`,
 and the sender silently receives nothing. Fail toward the recoverable error.
 
 ### Invariants — both roles, no exceptions
@@ -41,7 +41,7 @@ and the sender silently receives nothing. Fail toward the recoverable error.
    stays on subscription; only the bridge puts a member off it, at spawn. Mounting the bridge must
    never move a session across that boundary.
 2. **The bridge is the only channel.** Text you print in your terminal reaches nobody — the other
-   side cannot see your screen. An answer that isn't in a `bridge_*` call is silently discarded.
+   side cannot see your screen. An answer that isn't in a `fleet_*` call is silently discarded.
 3. **Identity comes from the connection, never an argument.** Workers never pass a target; you
    cannot act as another session. Spawn/stop/drain are lead-only; **send is lead or architect**;
    reply/ask are only-as-itself — any peer may answer for its own pane, and for no other. A call
@@ -57,10 +57,10 @@ and the sender silently receives nothing. Fail toward the recoverable error.
 
 **Delegate by default — that is the job.** With the bridge mounted you are an orchestrator on a
 metered subscription, and workers are cheap, parallel, and disposable. The default answer to "who
-does this?" is **a worker**, not you. Reach for `bridge_send` before you reach for `Edit`. The steps
+does this?" is **a worker**, not you. Reach for `fleet_send` before you reach for `Edit`. The steps
 below are the procedure — run them in order, every task, not only the big ones.
 
-0. **Know your role** — `bridge_whoami`, once per session, before anything else.
+0. **Know your role** — `fleet_whoami`, once per session, before anything else.
 1. **Split.** Write the unit list. Every unit carries: scope · the files or PR in question ·
    acceptance criteria · exactly what to report back. A unit with no acceptance criteria is not
    ready to delegate — refine it or keep it.
@@ -70,17 +70,17 @@ below are the procedure — run them in order, every task, not only the big ones
    delegate. The keep-list is closed: the conversation with the user, decomposition and planning,
    the final judgment call, verification, merges, and anything that depends on context only you
    hold. Nothing else is yours by default.
-3. **Spawn every delegated unit first** — `bridge_spawn{profile, worktree:true, ticket}`, one per
+3. **Spawn every delegated unit first** — `fleet_spawn{profile, worktree:true, ticket}`, one per
    unit, *before* sending any. Pass `profile` explicitly: profiles differ in model and cost, not in
    tier, so the default is rarely what you want.
-4. **Then send them all** — `bridge_send{sessionId, content, wait:false}`. Line 1 of every brief is
+4. **Then send them all** — `fleet_send{sessionId, content, wait:false}`. Line 1 of every brief is
    `Load the <name> skill.` naming the worker's playbook; those skills are opt-in and that line is
    what makes them reliable. Where the project ships no such skill, spell the procedure out in the
    brief instead. The brief is self-contained — the worker sees your message and the repo, nothing
    of your context, your plan, or your screen.
-5. **Collect** — `bridge_poll{ticket}` → `bridge_ack{target, msgId}`. Answer a worker's `bridge_ask`
-   with `bridge_send{turnId, content}` — **not** `sessionId`. A worker gone quiet is diagnosed with
-   `bridge_status`, never by reading its terminal; it also reports an open question and the `turnId`
+5. **Collect** — `fleet_poll{ticket}` → `fleet_ack{target, msgId}`. Answer a worker's `fleet_ask`
+   with `fleet_send{turnId, content}` — **not** `sessionId`. A worker gone quiet is diagnosed with
+   `fleet_status`, never by reading its terminal; it also reports an open question and the `turnId`
    that answers it. **A worker's ask waits ~55 seconds, and no nudge makes that longer** — so never
    brief a worker to "ask me". Decide before you delegate, or give it an explicit default.
 6. **Verify yourself.** Re-run the build and the checks. A worker cannot run your IDE tooling, any
@@ -93,11 +93,11 @@ below are the procedure — run them in order, every task, not only the big ones
    read it yourself.
 8. **Adjudicate, merge, tear down — yours alone.** Read the diff yourself: fully if it is small,
    targeted at the reported findings and the risky paths if it is large. Reviewer findings direct
-   your attention; they never substitute for it. Then merge, then `bridge_stop{paneId}`.
+   your attention; they never substitute for it. Then merge, then `fleet_stop{paneId}`.
 
 **Steps 3 and 4 are separate on purpose** — spawning and sending in one loop is how parallel work
 silently becomes serial, and it is the most common way this layer is wasted. For the same reason,
-prefer `wait:false` + `bridge_poll` for anything non-trivial: a blocking `bridge_send` is capped by
+prefer `wait:false` + `fleet_poll` for anything non-trivial: a blocking `fleet_send` is capped by
 *your own* MCP client call timeout (~60s), well below the task's real runtime.
 
 **Delegating does not delegate responsibility.** Workers open PRs; you are the gate. Never delegate
@@ -105,21 +105,22 @@ the merge — and merging on a reviewer's word is delegating it by proxy.
 
 | Intent | Tool |
 |---|---|
-| Confirm your own role | `bridge_whoami` |
-| See backends available | `bridge_profiles` |
-| Start a member | `bridge_spawn{role?, profile?, cwd?, worktree?, ticket?, sessionName?, resumeSessionId?}` → `sessionId` + `paneId` |
-| See the fleet | `bridge_list` → `leads` (your peers) + `members` (each carries `agentSessionId` when its backend knows one) · one peer's state: `bridge_status{sessionId}` |
-| Delegate (blocking) | `bridge_send{sessionId, content}` |
-| Delegate (long task) | `bridge_send{sessionId, content, wait:false}` → ticket → `bridge_poll{ticket}` |
-| Answer a member's `bridge_ask` | `bridge_send{turnId, content}` — **not** `sessionId` |
-| Message a **peer lead** | `bridge_send{sessionId: <their terminal>, content}` — `bridge_list` → `leads` reports it. Coordination only, **never** a task |
-| Answer a peer lead that messaged you | `bridge_reply{content}` — the one case a lead replies |
-| Collect a held reply | `bridge_poll{target}` · then `bridge_ack{target, msgId}` |
-| Tear down a member | `bridge_stop{paneId}` |
+| Confirm your own role | `fleet_whoami` |
+| See backends available | `fleet_profiles` |
+| Start a member | `fleet_spawn{role?, profile?, cwd?, worktree?, ticket?, sessionName?, resumeSessionId?}` → `sessionId` + `paneId` |
+| See the fleet | `fleet_list` → `leads` (your peers) + `members` (each carries `agentSessionId` when its backend knows one) · one peer's state: `fleet_status{sessionId}` |
+| Delegate (blocking) | `fleet_send{sessionId, content}` |
+| Delegate (long task) | `fleet_send{sessionId, content, wait:false}` → ticket → `fleet_poll{ticket}` |
+| Answer a member's `fleet_ask` | `fleet_send{turnId, content}` — **not** `sessionId` |
+| Message a **peer lead** on this host | `fleet_send{sessionId: <their terminal>, content}` — `fleet_list` → `leads` reports it. Coordination only, **never** a task |
+| Message a **peer lead** on another daemon or host | `fleet_send{coordId: <their coord-id>, content}` — needs a `coordinator:` block; your own coord-id is in `fleet_list`. Coordination only, **never** a task |
+| Answer a peer lead that messaged you | `fleet_reply{content}` — the one case a lead replies |
+| Collect a held reply | `fleet_poll{target}` · then `fleet_ack{target, msgId}` |
+| Tear down a member | `fleet_stop{paneId}` |
 
 ### Lead ↔ lead — coordinate, never delegate
 
-`bridge_list` returns `leads` alongside `members`; your own row carries `self: true`. Every other row
+`fleet_list` returns `leads` alongside `members`; your own row carries `self: true`. Every other row
 is a peer — an orchestrator with its own context, its own members, and its own judgment. An empty
 `members` array means no members are spawned; it says nothing about peers.
 
@@ -140,7 +141,7 @@ The traffic between leads is coordination and nothing else:
    against the code, and re-run the build. A peer's correction gets the same treatment — right or
    wrong on the evidence, not on who said it. Neither of you merges the other's work unreviewed.
 
-Being messaged by a peer does not make you its worker: answer with `bridge_reply`, and push back on
+Being messaged by a peer does not make you its worker: answer with `fleet_reply`, and push back on
 the substance if it is wrong. A peer that simply complies has thrown away the reason there are two of
 you.
 
@@ -148,11 +149,11 @@ you.
 
 1. **Load the playbook skill the lead named** before doing anything else.
 2. **Do the assigned scope only.** Note anything you spot outside it in one line; don't go hunt it.
-3. **`bridge_ask{question}`** when a decision is genuinely the lead's (ambiguous requirement, two
+3. **`fleet_ask{question}`** when a decision is genuinely the lead's (ambiguous requirement, two
    defensible fixes, "bug or intended?"). It blocks and you resume the *same* turn with the answer.
    Don't ask what you could decide yourself.
-4. **End the turn with exactly one `bridge_reply{content}`**, carrying your complete answer. This is
-   the whole handoff. No `bridge_reply` ⇒ the sender gets nothing and the exchange stalls.
+4. **End the turn with exactly one `fleet_reply{content}`**, carrying your complete answer. This is
+   the whole handoff. No `fleet_reply` ⇒ the sender gets nothing and the exchange stalls.
    Do **not** lean on the completion fallback to carry your answer for you: when you end a turn
    without replying, the bridge scrapes your pane, and it can return only the last 4000 characters.
    A clipped scrape is marked as partial, but the missing text is gone — your report reaches the
@@ -171,35 +172,61 @@ you.
 
 | Layer | Scope | Reaches |
 |---|---|---|
-| the launcher's reply charter | the one rule that must survive with no repo: *end every turn with `bridge_reply`* | every spawned member, at launch, every peer kind — never a lead |
+| the launcher's reply charter | the one rule that must survive with no repo: *end every turn with `fleet_reply`* | every spawned member, at launch, every peer kind — never a lead |
 | **this section** | protocol + orchestration policy | primary **and** every member that reads the repo — tracked in git, so worktrees inherit it |
+| role agent definition files | role contract and per-job procedure | a member whose launcher binds its role to the matching file in its worktree |
 | role playbook skills | per-job procedure (commit/PR recipe, finding format) | a member told to load one |
 | the bridge's own docs | design detail, flows, error model | on demand |
 
-A rule belongs in **exactly one** layer — the outermost one that must obey it. Peers that don't read
-`CLAUDE.md` (non-Claude adapters) get the charter only, so any rule *they* must obey belongs in the
-charter, not here.
+A rule belongs in **exactly one** layer — the outermost one that must obey it. A member without a
+repo checkout still gets the launcher's reply charter, which is why that one rule stays there.
+Peers that don't read `CLAUDE.md` (non-Claude adapters) get the charter only, so any rule *they*
+must obey belongs in the charter, not here.
 
 ## Project addendum — claude-bridge (not part of the canonical block)
 
-- **This repo is the bridge.** The daemon is `bridged`, its MCP mount is `http://127.0.0.1:8765/mcp`,
-  and the code behind the rules above is `mcp/BridgeMcp` (tools), `auth/Authz` (the role table),
+- **This repo is the bridge.** The daemon is `fleetd`, its MCP mount is `http://127.0.0.1:8765/mcp`,
+  and the code behind the rules above is `mcp/FleetMcp` (tools), `auth/Authz` (the role table),
   `mcp/ConnectionIdentity` (connection→role), and `worker/*Launcher` (`REPLY_CHARTER`).
+- **`fleet_profiles`/`fleet_list` report two separate outage states, and they are not the same
+  thing.** *Quarantined* (CB-578) means the backend told us it is out of capacity — a long,
+  1800s-default cooldown. *Cooling off* (fleetd #201/#227) means a profile's credential threw two
+  distinct backend errors (a non-exhaustion failure such as an HTTP 5xx) within 60 seconds — a
+  short, fixed 60s cooldown, not configurable per profile. Each check runs independently, so a
+  profile can show both at once. In the JSON: a cooling profile carries `credentialId` and
+  `coolingOffForSeconds`; a quarantined profile carries `quarantinedForSeconds`; a profile hit by
+  both carries all three fields, and either state alone already sets that profile's `free` to `0`.
+  A `fleet_spawn` naming a cooling-off profile is refused before it ever reaches the backend
+  adapter, with a message naming the credential and the remaining seconds ("cooling off after
+  repeated backend errors") — distinct wording from a quarantine refusal, so don't conflate the
+  two when reading a spawn failure.
 - **Skills available to delegate:** `implementer` (worktree → commit → push → own PR) and
   `reviewer` (scoped review → one structured finding). Name one in every delegation.
 - **Primary-side skills** (not delegation playbooks — a worker cannot use them):
-  `port-to-opencode` (make an OpenCode session a participant in this workspace).
+  `port-to-opencode` (make an OpenCode session a participant in this workspace) and
+  `fleets-status` (report every fleet that shares one LavinMQ instance).
 - **Never commit** `.mcp.json` (the primary's local copy, flagged `--skip-worktree`) or `wiki/`
   (a submodule with its own remote).
-- **Flows and the error model** — rendezvous, `bridge_ask`, detached delivery, turn-done fallback —
-  are diagrammed in `docs/MCP-Contract.md` **§6 only**. The rest of that page is a pre-build design
-  doc whose tool names, parameter names and REST paths never caught up with the code, so do not use
-  it as the tool reference (CB-609). Section 6 is kept out of this file because this file loads into
-  every session's context.
+- **A provisioned worktree neutralizes `.mcp.json`, `opencode.json` and `.autoenv`** — the repo's
+  committed copies would otherwise mount the primary's IDE and forge servers (fleetd #134). The
+  worktree's copy of each is a stub, **not** the repo's real file, so a worker that reads one and
+  reports what it found is reporting on the stub. The daemon logs a per-spawn summary, but the
+  worker cannot see that log. From inside its own worktree a worker — or a lead debugging one —
+  reads the list with `git config --worktree --get-all fleet.neutralizedConfig`, and the
+  consequence with `git config --worktree --get fleet.neutralizedConfigNote`. Never brief a worker
+  to edit one of these files: the edit cannot be committed, and it will not tell you so.
+- **Flows and the error model** — rendezvous, `fleet_ask`, detached delivery, the turn-done
+  fallback and status gating — are diagrammed in `docs/MCP-Contract.md`. That page is now flows
+  only: its pre-build tool catalogue, parameter tables and REST paths were deleted rather than
+  corrected, because a hand-maintained second copy of the tool surface is what drifted for a month
+  while this line pointed every session at it (CB-609 / #114). **The live MCP schema is the tool
+  reference**, with the intent→tool table above as the short form. `McpContractDocTest` fails if
+  that page names a `fleet_*` tool the server does not register. The flows are kept out of this
+  file because this file loads into every session's context.
 
 ### Redeploying the daemon — the lead may do this (primary only)
 
-**A merge is not a deployment.** The running `bridged` holds the jar it was started with, so a
+**A merge is not a deployment.** The running `fleetd` holds the jar it was started with, so a
 feature merged to `main` does nothing until the daemon is rebuilt and restarted. Saying "shipped"
 about code the live daemon has never loaded is a false report. The lead **may and should** redeploy
 rather than hand the job back to the operator.
@@ -210,9 +237,9 @@ and stopping it kills the worker's own channel mid-turn.
 **Use the script — do not hand-roll the steps.**
 
 ```bash
-scripts/redeploy-bridged.sh --check   # report state, change nothing
-scripts/redeploy-bridged.sh           # build, confirm drain, restart, verify
-scripts/redeploy-bridged.sh --yes     # skip the drain prompt (fleet already checked)
+scripts/redeploy-fleetd.sh --check   # report state, change nothing
+scripts/redeploy-fleetd.sh           # build, confirm drain, restart, verify
+scripts/redeploy-fleetd.sh --yes     # skip the drain prompt (fleet already checked)
 ```
 
 It builds before it stops anything, so a failed build never leaves the fleet down; it waits for the
@@ -229,17 +256,17 @@ if the script is unavailable or a step fails, this is what it was protecting you
    daemon starts fine, and the failure appears much later as workers that cannot open a PR. Nothing
    logs this at startup — the script's `--check` is the only thing that reports it, and it checks
    whether the name resolves without ever printing the value.
-2. **Drain live members first.** `bridge_list`, then `bridge_stop` each member, and collect anything
-   you still want with `bridge_poll` before you kill anything. A restart drops in-flight tickets and
+2. **Drain live members first.** `fleet_list`, then `fleet_stop` each member, and collect anything
+   you still want with `fleet_poll` before you kill anything. A restart drops in-flight tickets and
    rendezvous, and a member's report is not recoverable once its ticket is gone.
 3. **A restart is the only way deferred config keys take effect.** That is usually the reason to do
    it. The startup log names which keys it accepted and which it deferred — read those lines rather
    than assuming.
-4. **Re-check identity afterwards.** Call `bridge_whoami` and confirm it still answers `primary`. The
+4. **Re-check identity afterwards.** Call `fleet_whoami` and confirm it still answers `primary`. The
    lead is found by its tab label (`fleet.leaders.*.tab`), and a lead whose tab no longer matches is
    demoted to worker, which refuses every orchestration call.
-5. **Prove the new jar is the one running.** Confirm a *fresh* `bridged listening` line at the end of
-   `bridged/bridged.out`, dated after the restart. An old daemon that never died looks identical from
+5. **Prove the new jar is the one running.** Confirm a *fresh* `fleetd listening` line at the end of
+   `fleetd/fleetd.out`, dated after the restart. An old daemon that never died looks identical from
    the outside.
 
 **Permission.** A `CLAUDE.md` rule grants intent, not tool permission — the command classifier
@@ -248,7 +275,7 @@ it is one auditable command, so the operator allow-lists it once instead of appr
 start every time. The rule lives in the operator's Claude Code settings:
 
 ```json
-{ "permissions": { "allow": ["Bash(scripts/redeploy-bridged.sh:*)"] } }
+{ "permissions": { "allow": ["Bash(scripts/redeploy-fleetd.sh:*)"] } }
 ```
 
 Granted by the operator on 2026-08-15. If a call is still refused, do **not** route around it by
@@ -266,15 +293,15 @@ Before you call any work done, check the row that matches what you touched:
 
 | You changed… | Re-read and update… |
 |---|---|
-| a `bridge_*` tool — added, removed, renamed, or its params/semantics | the primary's intent→tool table; any rule that names that tool |
+| a `fleet_*` tool — added, removed, renamed, or its params/semantics | the primary's intent→tool table; any rule that names that tool |
 | `Authz` / the role table | invariant 3, and the primary-only vs worker-only claims |
-| `ConnectionIdentity` / how a caller is resolved | the `bridge_whoami` paragraph and the fallback ladder |
-| `REPLY_CHARTER`, or a launcher's mount/flags | the fallback ladder (`mcp__bridge__*`), and the layering table's top row |
+| `ConnectionIdentity` / how a caller is resolved | the `fleet_whoami` paragraph and the fallback ladder |
+| `REPLY_CHARTER`, or a launcher's mount/flags | the fallback ladder (`mcp__fleet__*`), and the layering table's top row |
 | the injector / status gating | invariant 4 |
 | worktree provisioning or the parity overlay | the "both roles read this file" premise — it rests on the worker's worktree being a checkout of this repo |
 | `.claude/skills/**` | the addendum's skill list, and the "name the playbook" rule |
 | a new peer kind (non-Claude adapter) | what that peer can read — anything it must obey belongs in its charter, not in the block |
-| **anything an operator can use, configure, or observe** — an MCP tool, a `bridged.yaml` knob, an endpoint, a visible behaviour | **[Features](wiki/11-Features.md)** — one entry: what it does · the knob that turns it on · **why it exists** · the gotcha |
+| **anything an operator can use, configure, or observe** — an MCP tool, a `fleetd.yaml` knob, an endpoint, a visible behaviour | **[Features](wiki/11-Features.md)** — one entry: what it does · the knob that turns it on · **why it exists** · the gotcha |
 
 That last row is not bookkeeping. Chapters 1–10 answer *how is this built* and *why this way*;
 none of them has a home for *what can it do and how do I turn it on*, so for twenty tickets a
@@ -285,7 +312,7 @@ is a Roadmap line. A change that touches none of the three earns no entry, and t
 outcome rather than an omission.
 
 Then **propagate**: the block in this file and the template in the wiki
-([Use Cases](https://git.ltms.dev/lms/claude-bridge/wiki/7-Use-Cases) → *The portable `CLAUDE.md`
+([Use Cases](https://git.ltms.dev/fleet/fleetd/wiki/7-Use-Cases) → *The portable `CLAUDE.md`
 block*) must stay byte-identical, and other projects carrying the block need the same edit. Verify
 rather than trust:
 
@@ -308,10 +335,10 @@ PY
 
 Two IDE MCP servers are connected: **intellij-index** (semantic code intelligence) and
 **jetbrains** (file problems, reformat, debugger). IntelliJ has multiple projects open; our
-module is **`bridged`**. Always pass these to IDE MCP tools:
+module is **`fleetd`**. Always pass these to IDE MCP tools:
 
-- `project_path` = `/Users/dai.ha/LTMS/claude-bridge/bridged`
-- IDE paths are relative to `bridged/` (e.g. `src/main/java/dev/ltms/bridged/...`)
+- `project_path` = `/Users/dai.ha/LTMS/claude-bridge/fleetd`
+- IDE paths are relative to `fleetd/` (e.g. `src/main/java/dev/ltms/fleet/...`)
 
 ### After editing any file — mandatory
 
@@ -325,7 +352,7 @@ module is **`bridged`**. Always pass these to IDE MCP tools:
    whole-project gate before declaring work done or committing.
 
 **Whenever dependencies change (or a `pom.xml` edit), validate CVEs with
-`jetbrains get_file_problems{filePath: "bridged/pom.xml"}`** — its Mend.io check reflects the
+`jetbrains get_file_problems{filePath: "fleetd/pom.xml"}`** — its Mend.io check reflects the
 dependencies on disk. (Note: `ide_diagnostics` / intellij-index does NOT re-resolve dependencies
 after a pom edit without a full Maven reimport, so it reports stale CVE results — don't trust it
 for this.) Treat a CVE warning like any other: bump to a patched version and confirm

README.md

@@ -9,23 +9,23 @@ Sibling of [`crush-bridge`](https://git.ltms.dev/systems/vms) (which drives a he
 process*, so it inherits `CLAUDE.md`, hooks, skills, and MCP — just pointed at a
 cheaper/local model.
 
-## Leading approach — herdr-centric message server (`bridged`)
+## Leading approach — herdr-centric message server (`fleetd`)
 
-A small always-on message server, **`bridged`**, controls
+A small always-on message server, **`fleetd`**, controls
 [herdr](https://herdr.dev) (an agent multiplexer) over its Unix-socket API and exposes a
 clean 2-way messaging API as an **MCP server that both the primary and the workers mount** —
 one unified Claude setup and the **sole communication gateway** (REST/SSE stays for non-Claude
-clients; any broker is `bridged`-internal, below the gateway).
-herdr owns the PTYs, multiplexing, persistence, and **agent-status events**; `bridged` owns
+clients; any broker is `fleetd`-internal, below the gateway).
+herdr owns the PTYs, multiplexing, persistence, and **agent-status events**; `fleetd` owns
 policy (subscription boundary, session lifecycle, status-gated delivery) and the client
 contract. A Claude member launches with `ANTHROPIC_BASE_URL` pointed at the gateway,
 `https://llm.ltms.dev/anthropic`, plus a bearer token; the lead stays env-clean and calls
-`bridged`'s MCP tools. See the wiki's **[13 User Guide](wiki/13-User-Guide.md)** to run it.
+`fleetd`'s MCP tools. See the wiki's **[13 User Guide](wiki/13-User-Guide.md)** to run it.
 
 ```mermaid
 flowchart LR
     OPUS["Opus — primary<br/>(Claude Code, env CLEAN)<br/>MCP client"]
-    subgraph BD["bridged — standalone daemon (not a claude process)"]
+    subgraph BD["fleetd — standalone daemon (not a claude process)"]
         SRV["SERVER face<br/>MCP · REST/SSE · policy"]
         CLI["CLIENT face<br/>status-gated injector · herdr socket"]
         SRV --> CLI
@@ -34,8 +34,8 @@ flowchart LR
     W["worker claude pane<br/>ANTHROPIC_BASE_URL set<br/>MCP client"]
     M["llm.ltms.dev<br/>(the one gateway)"]
 
-    OPUS -->|"MCP bridge_send (blocks)"| SRV
-    W -.->|"MCP bridge_reply"| SRV
+    OPUS -->|"MCP fleet_send (blocks)"| SRV
+    W -.->|"MCP fleet_reply"| SRV
     CLI -->|"Unix socket<br/>send_text · events.subscribe"| HERDR
     HERDR -->|"drives PTY"| W
     W -->|"inference"| M
@@ -47,24 +47,26 @@ flowchart LR
 ```
 
 - **Subscription boundary:** the *primary* never sets `ANTHROPIC_BASE_URL` (stays on
-  Pro/Max). Only the *secondary* process is off-subscription — and `bridged` itself is a
+  Pro/Max). Only the *secondary* process is off-subscription — and `fleetd` itself is a
   plain daemon (no Anthropic quota), so it may poll/subscribe freely.
-- **One gateway (unified MCP setup):** `bridged` is the **sole communication path** for every
+- **One gateway (unified MCP setup):** `fleetd` is the **sole communication path** for every
   Claude session. Primary and workers each mount it as an MCP server (one `claude mcp add`
-  line, same on both) and talk over MCP tools — `bridge_send` / `bridge_reply` /
-  `bridge_status` (with `bridge_ask` planned for the blocked-worker path). **No Claude session
-  ever addresses a broker, a peer, or the network
-  directly**; any queue is `bridged`-internal. MCP tool I/O never sets `ANTHROPIC_BASE_URL`, so
-  mounting the bridge is subscription-safe by construction.
-- **How the primary consumes a reply:** a single **blocking MCP call** (`bridge_send`);
-  `bridged` holds it open until the worker calls `bridge_reply` or its turn hits
+   line, same on both) and talk over MCP tools — `fleet_send` / `fleet_reply` /
+   `fleet_status` (with `fleet_ask` planned for the blocked-worker path). **No Claude session
+   ever addresses a broker, a peer, or the network
+   directly**; any queue is `fleetd`-internal. MCP tool I/O never sets `ANTHROPIC_BASE_URL`, so
+   mounting the bridge is subscription-safe by construction.
+   **Tool naming:** the tools are `fleet_*` (renamed from `bridge_*` in CB-622). The old
+   `bridge_*` names were removed in CB-634 — only `fleet_*` answers now.
+- **How the primary consumes a reply:** a single **blocking MCP call** (`fleet_send`);
+  `fleetd` holds it open until the worker calls `fleet_reply` or its turn hits
   `agent_status=done`, then returns the reply as the tool result. No cross-turn busy-poll, so
   no quota burn. SSE is an optional side-channel for humans/dashboards watching status.
-- **Worker → primary** rides `bridged`'s **MCP rendezvous** — the reply resolves the primary's
-  blocking call (or, for detached work, `bridged` **injects the primary's idle pane** when it's
+- **Worker → primary** rides `fleetd`'s **MCP rendezvous** — the reply resolves the primary's
+  blocking call (or, for detached work, `fleetd` **injects the primary's idle pane** when it's
   ready), so *no keystroke-into-primary and no broker are involved, even single-host*. The one
   exception: a split-host primary that isn't a herdr pane wakes via its own `Stop`-hook, which
-  polls **`bridged`** (never a broker). See the wiki for the two topologies.
+  polls **`fleetd`** (never a broker). See the wiki for the two topologies.
 - **Different model per process** sidesteps Claude Code's lack of per-subagent provider
   routing — the worker isn't a subagent, it's its own configured process.
 - **AgentAPI** ([`coder/agentapi`](https://github.com/coder/agentapi)) is retained only as a
@@ -73,11 +75,11 @@ flowchart LR
 
 ## Docs
 
-Full design, setup, and operations live in the **[wiki](https://git.ltms.dev/lms/claude-bridge/wiki)**,
+Full design, setup, and operations live in the **[wiki](https://git.ltms.dev/fleet/fleetd/wiki)**,
 vendored here as a submodule under [`wiki/`](./wiki):
 
 ```bash
-git clone --recurse-submodules ssh://git@git.ltms.dev:2224/lms/claude-bridge.git
+git clone --recurse-submodules ssh://git@git.ltms.dev:2224/fleet/fleetd.git
 # or, after a plain clone:
 git submodule update --init
 ```
@@ -87,7 +89,7 @@ Gitea wiki.
 
 ## Status
 
-🟢 **Implemented & dogfooded** — the herdr-centric **`bridged`** message server is built and in
+🟢 **Implemented & dogfooded** — the herdr-centric **`fleetd`** message server is built and in
 real use: an Opus primary delegates tasks to off-subscription workers that reply through the
 bridge (code reviews delegated this way have produced committed bug fixes). Selected as the
 primary approach 2026-07-11, superseding the AgentAPI plan (2026-07-08); AgentAPI retained as a
@@ -98,9 +100,9 @@ tests run separately via `mvn test -Pcontract`):
 
 - **Core gateway** — herdr socket client (contract-tested vs live 0.7.0); guard-checked worker
   spawn with `ANTHROPIC_BASE_URL` injected only into the worker's env; status-gated injector;
-  blocking `bridge_send` with reply rendezvous; MCP server as a thin adapter over the REST core.
-- **MCP tools** — `bridge_send` / `bridge_reply` / `bridge_status` (messaging) and `bridge_spawn`
-  / `bridge_list` / `bridge_stop` / `bridge_profiles` / `bridge_poll` (fleet). Caller identity is
+  blocking `fleet_send` with reply rendezvous; MCP server as a thin adapter over the REST core.
+- **MCP tools** — `fleet_send` / `fleet_reply` / `fleet_status` (messaging) and `fleet_spawn`
+  / `fleet_list` / `fleet_stop` / `fleet_profiles` / `fleet_poll` (fleet). Caller identity is
   connection-based (loopback peer PID → herdr pane), so the same mount serves primary and workers.
 - **Delivery reliability** — completion fallback (a confirmed `working→idle` turn resolves a
   send); async fire-and-poll (beats the caller's MCP call timeout for long tasks); and failure
@@ -108,7 +110,7 @@ tests run separately via `mvn test -Pcontract`):
 - **Fleet** — multiple worker profiles, each with an independent base_url guard check; workers
   inherit the primary's working directory (never `$HOME`); a readiness gate holds delivery until
   a worker's Claude has connected the bridge MCP (no paste lost into its boot window).
-- **Blocked-worker path** — `bridge_ask` reverse rendezvous: a worker pauses its delegated turn to
+- **Blocked-worker path** — `fleet_ask` reverse rendezvous: a worker pauses its delegated turn to
   ask the primary and resumes the *same* turn with the answer (CB-205).
 - **Session lifecycle** — session manager with spawn/reuse/recycle, `idle_ttl` reaper, `context_cap`,
   and graceful drain on shutdown (CB-301/CB-303); per-worker git worktrees on their own branch with

bridged/.gitignore

@@ -1,14 +0,0 @@
-# Build output
-target/
-dependency-reduced-pom.xml
-
-# Local runtime config (copy from bridged.example.yaml)
-bridged.yaml
-
-# CB-505 audit trail + daemon stdout/stderr — runtime records, never source
-logs/
-
-# Editor / OS
-*.iml
-.idea/
-.DS_Store

bridged/src/main/java/dev/ltms/bridged/Bridged.java

@@ -1,683 +0,0 @@
-package dev.ltms.bridged;
-
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.config.ConfigRef;
-import dev.ltms.bridged.config.ConfigWatcher;
-import dev.ltms.bridged.guard.SubscriptionGuard;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.HerdrClient;
-import dev.ltms.bridged.herdr.HerdrException;
-import dev.ltms.bridged.herdr.LeadTabScanner;
-import dev.ltms.bridged.lead.LeadLauncher;
-import dev.ltms.bridged.herdr.PaneLocator;
-import dev.ltms.bridged.herdr.UnixSocketHerdrClient;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.inject.CompletionResolver;
-import dev.ltms.bridged.inject.ExhaustedPatternLookup;
-import dev.ltms.bridged.inject.ExhaustionSink;
-import dev.ltms.bridged.inject.Injector;
-import dev.ltms.bridged.inject.StatusPoller;
-import dev.ltms.bridged.inject.TurnListener;
-import dev.ltms.bridged.inject.MemberPresence;
-import dev.ltms.bridged.auth.MemberRegistry;
-import dev.ltms.bridged.auth.CallerResolver;
-import dev.ltms.bridged.mcp.BridgeMcp;
-import dev.ltms.bridged.mcp.ConnectionIdentity;
-import dev.ltms.bridged.metrics.BridgedMetrics;
-import dev.ltms.bridged.metrics.Metrics;
-import dev.ltms.bridged.health.FleetHealthMonitor;
-import dev.ltms.bridged.mcp.PrimaryRegistry;
-import dev.ltms.bridged.mcp.LsofPeerPidLookup;
-import dev.ltms.bridged.mcp.LsofProcessCwdLookup;
-import dev.ltms.bridged.msg.AmqpReplyInbox;
-import dev.ltms.bridged.msg.InMemoryReplyInbox;
-import dev.ltms.bridged.msg.MessageService;
-import dev.ltms.bridged.msg.Rendezvous;
-import dev.ltms.bridged.msg.ReplyInbox;
-import dev.ltms.bridged.msg.LeadHeartbeatLoop;
-import dev.ltms.bridged.msg.ReplyPushLoop;
-import dev.ltms.bridged.rest.BridgedApp;
-import dev.ltms.bridged.session.GitWorktrees;
-import dev.ltms.bridged.session.MemberSession;
-import dev.ltms.bridged.session.SessionManager;
-import dev.ltms.bridged.peer.PeerLauncher;
-import dev.ltms.bridged.session.SessionReaper;
-import dev.ltms.bridged.member.ClaudeCodeLauncher;
-import dev.ltms.bridged.member.CompositePeerLauncher;
-import dev.ltms.bridged.member.HerdrPeerLauncher;
-import dev.ltms.bridged.member.OpenCodeLauncher;
-import dev.ltms.bridged.placement.BackendQuarantine;
-import io.javalin.Javalin;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-
-import java.nio.file.Path;
-import java.util.ArrayList;
-import java.util.LinkedHashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Objects;
-import java.util.Set;
-import java.util.concurrent.Executors;
-import java.util.concurrent.TimeUnit;
-import java.util.concurrent.atomic.AtomicReference;
-import java.util.function.Function;
-import java.util.function.Predicate;
-import java.util.function.Supplier;
-import java.util.regex.Pattern;
-import java.util.stream.Collectors;
-
-/**
- * {@code bridged} entry point. Wires the real herdr socket client to the REST app and
- * starts listening. Before anything else it asserts its own environment is clean —
- * {@code bridged} is not a Claude process and must never carry a base_url.
- */
-public final class Bridged {
-
-    private static final Logger log = LoggerFactory.getLogger(Bridged.class);
-
-    /** CB-504: how long to wait at startup for herdr's socket before serving degraded. */
-    private static final long HERDR_WAIT_SECONDS = 30;
-    private static final long HERDR_WAIT_POLL_MILLIS = 500;
-
-    static void main(String[] args) {
-        Path configPath = Path.of(args.length > 0 ? args[0] : "bridged.yaml");
-        BridgedConfig cfg = BridgedConfig.load(configPath);
-        // CB-594: report which secret env vars the config actually needs, by name, before anything
-        // else can fail on a silently-empty one. A daemon started without a login shell (launchd)
-        // boots fine either way — this is the only thing that says so out loud.
-        reportRequiredSecrets(cfg);
-        // CB-596: an absent (or empty) memberCredentials: block blocks NOTHING — no credential
-        // name is hardcoded any more to fall back on. Say so loudly, the same way a missing
-        // secret is reported above, so upgrading past this commit never silently drops CB-592's
-        // protection.
-        reportMemberCredentialsGap(cfg);
-        // CB-559: `cfg` stays the startup snapshot — every validation and every piece of one-time
-        // wiring below reads it, and must, because those decisions cannot be unmade. `config` is the
-        // live reference the hot paths read per use. Which keys can actually move is ConfigRef's
-        // contract; adding a reader here does not make a key reloadable by itself.
-        ConfigRef config = new ConfigRef(configPath, cfg);
-
-        // The primary/host env that launched bridged must not be tainted.
-        SubscriptionGuard guard = new SubscriptionGuard(cfg.guard().hostSet());
-        guard.assertPrimaryClean(System.getenv());
-
-        // CB-501: refuse to start if the bind is wider than the auth mode can defend. Under
-        // loopback-trust, "not a known worker" means "the primary" — sound only because the OS
-        // refuses remote connections to a loopback socket. This throws rather than warns so the
-        // dangerous configuration cannot be reached by ignoring a log line.
-        cfg.validateAuthExposure();
-        cfg.validateLeadTabPrefixes();
-        // CB-542: a subscription:true profile whose env: reseats ANTHROPIC_BASE_URL/AUTH_TOKEN would
-        // reach an unguarded endpoint (the launcher skips SubscriptionGuard for it). Refuse at load.
-        cfg.validateSubscriptionProfiles();
-        cfg.validateCharters();
-        // CB-548: every architect slot must name a configured workers: profile — the strong-model
-        // backend the future spawn lifecycle would read. A stale reference dies here, not later.
-        cfg.validateMembers();
-
-        Path socket = cfg.herdrSocket() != null && !cfg.herdrSocket().isBlank()
-                ? Path.of(cfg.herdrSocket())
-                : UnixSocketHerdrClient.defaultSocketPath();
-
-        UnixSocketHerdrClient herdr = UnixSocketHerdrClient.connect(socket, new com.fasterxml.jackson.databind.ObjectMapper());
-
-        AgentControl agents = new AgentControl(herdr);
-        WorkspaceControl spaces = new WorkspaceControl(herdr);
-        // CB-402: one adapter per configured peer kind, fronted by a composite router. A profile's
-        // `kind:` selects its adapter — claude-code (the default) and opencode partition the profile
-        // set — and the composite dispatches each SPI call to the adapter that owns the profile/pane.
-        Map<String, BridgedConfig.Profile> claudeProfiles = new LinkedHashMap<>();
-        Map<String, BridgedConfig.Profile> opencodeProfiles = new LinkedHashMap<>();
-        cfg.profiles().forEach((name, w) -> {
-            if (w.isOpenCode()) {
-                opencodeProfiles.put(name, w);
-            } else {
-                claudeProfiles.put(name, w);
-            }
-        });
-        List<HerdrPeerLauncher> adapters = new ArrayList<>();
-        // The claude-code adapter is the always-present default; keep it even with no profiles (so a
-        // bridge configured with no workers, or opencode-only, still has a well-defined base adapter)
-        // unless opencode is the only kind configured.
-        if (!claudeProfiles.isEmpty() || opencodeProfiles.isEmpty()) {
-            adapters.add(new ClaudeCodeLauncher(agents, spaces, guard,
-                    claudeProfiles, cfg.effectiveDefaultProfile(), System::getenv,
-                    cfg.spawnReadyTimeoutMs(), cfg.spawnReadyPollMs(),
-                    () -> config.get().fleet(),
-                    () -> config.get().memberCredentials()));
-        }
-        if (!opencodeProfiles.isEmpty()) {
-            adapters.add(new OpenCodeLauncher(agents, spaces,
-                    opencodeProfiles, cfg.effectiveDefaultProfile(), System::getenv,
-                    cfg.spawnReadyTimeoutMs(), cfg.spawnReadyPollMs(),
-                    () -> config.get().fleet(),
-                    () -> config.get().memberCredentials()));
-        }
-        AtomicReference<Function<String, Integer>> liveCountRef = new AtomicReference<>(_ -> 0);
-        // CB-578 stage B: one quarantine tracker for the whole daemon, shared between the launcher
-        // (checked at spawn) and the exhaustion sink wired in below (written on BACKEND_EXHAUSTED).
-        // The cooldown is deferred (see BridgedConfig#quarantineCooldownSeconds): it is read once
-        // here, at startup, and a config reload only changes it for a daemon restart.
-        BackendQuarantine quarantine = new BackendQuarantine(System::nanoTime,
-                TimeUnit.SECONDS.toNanos(cfg.quarantineCooldownSeconds()));
-        PeerLauncher workers = new CompositePeerLauncher(
-                adapters,
-                cfg.effectiveDefaultProfile(),
-                config,
-                profileName -> liveCountRef.get().apply(profileName),
-                quarantine);
-        // CB-504: under supervision (launchd/systemd) bridged can start before herdr's socket
-        // exists. The client itself is lazy — it connects per call — but the orphan reap below is
-        // the first thing that actually talks to herdr, so without this wait a boot-order race
-        // would crash the daemon into a restart loop. Wait, then degrade rather than die: serving
-        // with /healthz reporting "degraded" is strictly more useful than exiting.
-        boolean herdrUp = awaitHerdr(herdr);
-        if (herdrUp) {
-            // CB-117: herdr keeps worker panes alive across a daemon restart, and their ids died
-            // with the previous process — reap those leaked orphans now, before we start serving.
-            workers.reapOrphanWorkers();
-        } else {
-            log.warn("herdr did not answer within {}s — starting anyway; /healthz will report "
-                    + "degraded until it comes up. Orphaned worker panes (if any) were NOT reaped.",
-                    HERDR_WAIT_SECONDS);
-        }
-
-        // CB-301: authoritative session registry + lifecycle FSM on top of ClaudeCodeLauncher.
-        // CB-301-ext: worktree provisioning seam, optionally rooted at a configured directory.
-        // CB-303 part 2: context cap is opt-in and disabled (0) when absent/null.
-        int contextCap = 0;
-        if (cfg.lifecycle() != null && cfg.lifecycle().contextCap() != null
-                && cfg.lifecycle().contextCap() > 0) {
-            contextCap = cfg.lifecycle().contextCap();
-        }
-        boolean clearAfterTurn = cfg.lifecycle() != null && cfg.lifecycle().clearAfterTurn();
-        SessionManager sessions = new SessionManager(workers, new GitWorktrees(cfg.worktreeRoot()),
-                System::nanoTime, contextCap, clearAfterTurn);
-        liveCountRef.set(profileName -> (int) sessions.roster().stream()
-                .filter(s -> profileName.equals(s.profile()))
-                .count());
-
-        // CB-303 part 1: idle-ttl reaper — only when configured, defaults to disabled.
-        final SessionReaper reaper;
-        if (cfg.lifecycle() != null
-                && cfg.lifecycle().idleTtlSeconds() != null
-                && cfg.lifecycle().idleTtlSeconds() > 0) {
-            reaper = new SessionReaper(sessions, cfg.lifecycle().idleTtlSeconds());
-            reaper.start();
-        } else {
-            reaper = null;
-        }
-
-        // CB-530: every pane the config names as a lead, merged from `leaders:` and the legacy
-        // singular pin. PrimaryRegistry below still tracks ONE terminal — it addresses the push
-        // loop's nudges, which need a single destination — so it keeps the legacy pin.
-        Map<String, String> leadTerminals = cfg.leaderTerminals();
-        if (leadTerminals.size() > 1) {
-            log.info("leads: {} panes recognised {}", leadTerminals.size(), leadTerminals.values());
-        }
-        // CB-531: on top of the legacy primary.terminal pin, discover leads by the tab labels the
-        // operator writes. CB-557 moved the settings onto the lead they describe, so scanning is on
-        // whenever a `fleet.leaders:` entry exists — with no leads configured the supplier is a
-        // constant and never touches herdr, exactly as a missing `leadScan:` block used to behave.
-        // CB-579: each lead now names its own exact `tab:` label, so one scanner discovers every
-        // configured lead regardless of how differently their tabs are labelled — the old
-        // single-shared-tabPrefix limitation (and its warning) is gone.
-        final Supplier<Map<String, String>> leads;
-        var leaders = cfg.fleet().leaders();
-        if (!leaders.isEmpty()) {
-            Set<String> memberSpaces = cfg.profiles().values().stream()
-                    .map(BridgedConfig.Profile::workspace)
-                    .filter(Objects::nonNull)
-                    .collect(Collectors.toSet());
-            Map<String, String> tabToName = new LinkedHashMap<>();
-            leaders.forEach((name, leader) -> {
-                if (leader != null && leader.tab() != null && !leader.tab().isBlank()) {
-                    tabToName.put(leader.tab(), name);
-                }
-            });
-            // One shared rescan cadence: still taken from the first entry, as before — it is an
-            // operational cadence, not identity, so there is no correctness reason to give every
-            // lead its own scanner.
-            int scanIntervalSeconds = leaders.values().iterator().next().scanIntervalSeconds();
-            leads = new LeadTabScanner(herdr, tabToName, memberSpaces,
-                    TimeUnit.SECONDS.toNanos(scanIntervalSeconds), System::nanoTime);
-            log.info("lead scan: tabs {} host a lead (rescan every {}s, member spaces {} excluded)",
-                    tabToName.keySet(), scanIntervalSeconds, memberSpaces);
-        } else {
-            leads = () -> leadTerminals;
-        }
-
-        // CB-558: start any declared lead that is not already running. After the scanner is built,
-        // because both read the same tab labels and the ordering makes that dependency visible; and
-        // only when herdr answered, because the launcher's whole safety property is that it can
-        // count live leads first — it must never guess and risk a second orchestrator.
-        if (herdrUp && !leaders.isEmpty()) {
-            int launched = new LeadLauncher(agents, spaces, cfg).ensureLeads();
-            if (launched > 0) {
-                log.info("lead auto-launch: {} lead(s) started", launched);
-            }
-        }
-
-        // CB-548: config-declared architect slots. Config supplies only the stable name → profile
-        // map; the terminal → slot binding is owned by the registry and is empty at startup, so no
-        // pane resolves to an architect until the later spawn lifecycle binds one. The registry is
-        // what CallerResolver resolves against and what that lifecycle will read profiles from;
-        // nothing here spawns a slot.
-        MemberRegistry members = new MemberRegistry(cfg.fleet());
-        sessions.setMemberLifecycle(members);
-        if (!members.slots().isEmpty()) {
-            log.info("member slots: {} configured {} — none bound yet (a slot is idle until the "
-                            + "spawn lifecycle binds a live terminal to it)",
-                    members.slots().size(), members.slots().keySet());
-        }
-
-        // Status-gated injector (CB-103): the single writer into workers, fed by a poller.
-        // The blocking message endpoint (CB-104) is the producer; the poller is inert until then.
-        // CB-106: a confirmed turn completion resolves a blocked send whose worker never replied.
-        Rendezvous rendezvous = new Rendezvous();
-        // CB-578 stage A: classify a completion-fallback scrape that matches a profile's configured
-        // usage-limit refusal as BACKEND_EXHAUSTED rather than handing it back as a real answer.
-        // Compiled once at startup, keyed by profile name; a profile with no exhaustedPattern is
-        // simply absent here, so its workers keep today's completion-fallback behaviour unchanged.
-        Map<String, Pattern> exhaustedPatternsByProfile = new LinkedHashMap<>();
-        cfg.profiles().forEach((name, profile) -> {
-            if (profile.hasExhaustedPattern()) {
-                exhaustedPatternsByProfile.put(name, Pattern.compile(profile.exhaustedPattern()));
-            }
-        });
-        ExhaustedPatternLookup exhaustedPatterns = target -> sessions.roster().stream()
-                .filter(session -> target.equals(session.terminalId()))
-                .findFirst()
-                .map(session -> exhaustedPatternsByProfile.get(session.profile()))
-                .orElse(null);
-        log.info("backend-exhausted classification (CB-578 stage A): {}",
-                CompletionResolver.coverage(cfg.profiles().keySet(), exhaustedPatternsByProfile.keySet()));
-        // CB-578 stage B: on a classification that actually wins, quarantine the exhausted profile's
-        // CREDENTIAL — not the profile name — so a profile sharing that credential (e.g. two models
-        // on one OpenAI account) is refused too, not just the one that happened to report it. Reads
-        // the profile config live off `config`, so a credentialId edit is hot: no restart needed.
-        ExhaustionSink exhaustionSink = (target, reason) -> sessions.roster().stream()
-                .filter(session -> target.equals(session.terminalId()))
-                .findFirst()
-                .map(MemberSession::profile)
-                .map(profileName -> config.get().profiles().get(profileName))
-                .ifPresent(profile -> {
-                    String credentialId = profile.effectiveCredentialId();
-                    quarantine.quarantine(credentialId);
-                    log.warn("credential '{}' quarantined for {}s (profile '{}' classified "
-                                    + "BACKEND_EXHAUSTED): {}", credentialId,
-                            cfg.quarantineCooldownSeconds(), profile.profile(), reason);
-                });
-        CompletionResolver completion = new CompletionResolver(agents, rendezvous, exhaustedPatterns, exhaustionSink);
-        // CB-113: deliver only to an available worker (its MCP is connected), never its boot window.
-        // CB-301: the manager's presence bridge records availability and drives SPAWNING → READY.
-        MemberPresence presence = sessions.asPresence();
-        TurnListener turnListener = new TurnListener() {
-            @Override
-            public void onTurnComplete(String target) {
-                completion.onTurnComplete(target);
-                sessions.onTurnComplete(target);
-            }
-
-            @Override
-            public boolean hasPostTurnAction(String target) {
-                return sessions.hasPostTurnAction(target);
-            }
-
-            @Override
-            public boolean onTurnCompleteWithPostAction(String target) {
-                completion.resolveBeforePostAction(target);
-                return sessions.onTurnCompleteWithPostAction(target);
-            }
-
-            @Override
-            public void onDelivered(String target, dev.ltms.bridged.msg.TurnToken token) {
-                completion.onDelivered(target, token);
-                sessions.onDelivered(target, token);
-            }
-
-            @Override
-            public void onTurnFailed(String target) {
-                completion.onTurnFailed(target);
-                sessions.onTurnFailed(target);
-            }
-
-            @Override
-            public void onTurnFailed(String target, String reason) {
-                completion.onTurnFailed(target, reason);
-                sessions.onTurnFailed(target);
-            }
-        };
-        Injector injector = new Injector(agents, turnListener, deliverableTo(presence, leads),
-                presence::forget);
-        StatusPoller poller = new StatusPoller(agents, injector, Injector.POLL_INTERVAL_MILLIS);
-        poller.start();
-
-        // CB-307: reply inbox. A broker: block (with a uri) selects the AMQP-backed durable adapter;
-        // absent, bridged stays soft-state on the in-memory inbox. The AMQP inbox owns a broker
-        // connection, so keep the reference to close it in the ordered shutdown hook.
-        final ReplyInbox replyInbox;
-        if (cfg.broker() != null && cfg.broker().isConfigured()) {
-            replyInbox = AmqpReplyInbox.open(cfg.broker().uri(), cfg.broker().prefetchOrDefault());
-            log.info("reply inbox: AMQP broker (durable) at {} (prefetch={})",
-                    cfg.broker().uri(), cfg.broker().prefetchOrDefault());
-        } else {
-            replyInbox = new InMemoryReplyInbox();
-            log.info("reply inbox: in-memory (soft-state)");
-        }
-        // CB-307: learn the primary's terminal from orchestration tool calls (or pin from config).
-        // The pin also feeds CallerResolver below: a primary running inside a herdr pane would
-        // otherwise resolve as a worker and be refused every orchestration tool.
-        String pinnedPrimaryTerminal = cfg.primary() != null ? cfg.primary().terminal() : null;
-        PrimaryRegistry primaryRegistry = new PrimaryRegistry(pinnedPrimaryTerminal);
-        // CB-532: `primary.terminal` is superseded and no longer needed for either of its jobs —
-        // identity comes from `leaders:`/`leadScan:`, and reply nudges now follow the delegating
-        // lead. Say so once at startup rather than leaving a redundant pin to look load-bearing.
-        if (pinnedPrimaryTerminal != null && !pinnedPrimaryTerminal.isBlank()) {
-            log.warn("primary.terminal is DEPRECATED (CB-532) and can be deleted: identity now comes "
-                    + "from leaders:/leadScan:, and reply nudges follow the lead that delegated. "
-                    + "It still works, and is still the fallback nudge destination when a restart "
-                    + "has lost the delegation map. Its pushReminders/pushBackoffMs stay valid.");
-        }
-        // CB-307: active push-to-primary loop — nudge the primary when replies land without an
-        // open bridge_send. Uses its own lightweight scheduled executor, separate from the injector.
-        int maxReminders = cfg.primary() != null ? cfg.primary().remindersOrDefault() : 5;
-        long backoffMs = cfg.primary() != null ? cfg.primary().backoffMsOrDefault() : 15_000L;
-        var pushScheduler = Executors.newSingleThreadScheduledExecutor(r ->
-                Thread.ofVirtual().name("bridge-push-").unstarted(r));
-        // CB-502: the registry is built before the service and the push loop so send/reply outcomes
-        // are counted at their single funnel rather than at each of the two caller-facing surfaces.
-        // CB-512: the push loop takes it too, so nudge outcomes (delivered|exhausted) are counted.
-        Metrics metrics = BridgedMetrics.create(sessions, replyInbox);
-        var pushLoop = new ReplyPushLoop(primaryRegistry, agents, replyInbox,
-                pushScheduler, maxReminders, backoffMs, metrics);
-        // CB-551: idle-lead heartbeat. Opt-in; absent `leadHeartbeat:` this is never constructed, so
-        // an upgraded daemon cannot silently start spending subscription on nudging an idle lead.
-        // It has its own single-thread scheduler and holds its own scheduler shutdown via close().
-        final LeadHeartbeatLoop heartbeat;
-        var heartbeatScheduler = Executors.newSingleThreadScheduledExecutor(r ->
-                Thread.ofVirtual().name("bridge-heartbeat-").unstarted(r));
-        if (cfg.leadHeartbeat() != null) {
-            var hb = cfg.leadHeartbeat();
-            heartbeat = new LeadHeartbeatLoop(primaryRegistry, agents, replyInbox, sessions::roster,
-                    pushLoop, heartbeatScheduler, System::nanoTime,
-                    TimeUnit.SECONDS.toNanos(hb.idleAfterSeconds()), hb.backoffMs(), hb.quietNudgeCap(),
-                    metrics);
-            heartbeat.start();
-        } else {
-            heartbeat = null;
-            heartbeatScheduler.shutdownNow();
-        }
-        MessageService messages = new MessageService(agents, injector, rendezvous, replyInbox,
-                pushLoop, metrics);
-
-        // Health is a slow whole-fleet observer. Keep it separate from the 250ms delivery poller.
-        final FleetHealthMonitor healthMonitor;
-        var healthScheduler = Executors.newSingleThreadScheduledExecutor(r ->
-                Thread.ofVirtual().name("bridge-health-").unstarted(r));
-        if (cfg.health() != null && cfg.health().isEnabled()) {
-            // CB-580: a member found GONE/NEVER_READY must fail whatever ticket is waiting on it,
-            // through the same idempotent target-wide operation CB-516 already uses on release.
-            healthMonitor = new FleetHealthMonitor(agents, sessions::roster, messages, healthScheduler,
-                    System::nanoTime, cfg.health().intervalOrDefault(), messages::abandon);
-            String coverage = FleetHealthMonitor.coverage(true,
-                    cfg.health().notifications() != null && cfg.health().notifications().configured());
-            if ("detection-only".equals(coverage)) {
-                log.warn("fleet health: {} (no notification sink configured)", coverage);
-            } else {
-                log.info("fleet health: {}", coverage);
-            }
-            healthMonitor.start();
-        } else {
-            healthMonitor = null;
-            healthScheduler.shutdownNow();
-        }
-
-        // CB-520: the reply inbox only consumes for agents this gateway owns. own on acquire,
-        // release on teardown. Do this before CB-516 so the inbox is owned before any reply can land.
-        sessions.onAcquire(replyInbox::own);
-        // CB-516: releasing a worker must fail whatever send was waiting on it. Without this a
-        // torn-down delegation kept reporting PENDING until the 30-minute async timeout, and never
-        // reached /metrics — the delegation was unresolvable and nothing said so.
-        sessions.onRelease(detail -> {
-            // CB-578 stage C, acceptance criterion 10: a failed ticket's detail should tell a lead
-            // where to re-dispatch onto the same tree, not just that the worker vanished.
-            String reason = "the worker session was released before it replied";
-            if (detail.worktreePath() != null) {
-                reason += "; worktree=" + detail.worktreePath() + " branch=" + detail.branch()
-                        + " snapshot=" + (detail.snapshotRef() != null ? detail.snapshotRef() : "none");
-            }
-            // CB-584 (issue #65 criterion 5): also name the agent session, so a lead can resume the
-            // member's conversation instead of only re-dispatching a fresh one onto the same files.
-            if (detail.agentSessionId() != null) {
-                reason += " agentSessionId=" + detail.agentSessionId();
-            }
-            messages.abandon(detail.terminalId(), reason);
-            replyInbox.release(detail.terminalId());
-            primaryRegistry.forgetDelegation(detail.terminalId()); // CB-532: don't leak the lead binding
-        });
-
-        // MCP server face (CB-105): bridge_send/bridge_reply/bridge_status, mounted at /mcp.
-        // Caller identity is resolved from the connection (peer PID → herdr pane), not arguments.
-        ConnectionIdentity identity = new ConnectionIdentity(
-                new PaneLocator(herdr), new LsofPeerPidLookup(), new LsofProcessCwdLookup());
-
-        // CB-501: one resolver behind both entry paths. Worker identity still comes from the
-        // connection and is never token-gated, so enabling token mode cannot lock the fleet out.
-        final CallerResolver callers;
-        if (cfg.auth().tokenMode()) {
-            String token = System.getenv(cfg.auth().tokenEnv());
-            if (token == null || token.isBlank()) {
-                throw new IllegalStateException("auth.mode=token but env var " + cfg.auth().tokenEnv()
-                        + " is unset or empty — export it before starting bridged");
-            }
-            callers = CallerResolver.withLeadsAndMembers(identity, true, token, leads, members);
-            log.info("auth: token mode (bearer required for non-worker callers, env {})",
-                    cfg.auth().tokenEnv());
-        } else {
-            callers = CallerResolver.withLeadsAndMembers(identity, false, null, leads, members);
-            log.info("auth: loopback-trust (any loopback non-worker caller is the primary)");
-        }
-
-        BridgeMcp mcp = new BridgeMcp(messages, workers, sessions, identity, presence,
-                primaryRegistry, callers, metrics, new BridgeMcp.CapacitySource(profile -> liveCountRef.get().apply(profile),
-                profile -> {
-                    var configured = config.get().profiles().get(profile);
-                    return configured == null ? null : configured.maxLoad();
-                }, () -> config.get().profiles().keySet(), System::nanoTime),
-                new BridgeMcp.HealthCoverageSource(() -> {
-                    var health = config.get().health();
-                    return FleetHealthMonitor.coverage(health != null && health.isEnabled(),
-                            health != null && health.notifications() != null && health.notifications().configured());
-                }),
-                new BridgeMcp.QuarantineSource(profile -> {
-                    var configured = config.get().profiles().get(profile);
-                    return configured == null ? null : configured.effectiveCredentialId();
-                }, quarantine));
-
-        // CB-559: opt-in config reload. With no `configReload:` block nothing is constructed, so an
-        // upgraded daemon behaves exactly as before — the file is read once at boot and never again.
-        final ConfigWatcher configWatcher;
-        if (cfg.configReload() != null && cfg.configReload().isEnabled()) {
-            configWatcher = new ConfigWatcher(config, cfg.configReload().intervalSeconds());
-            configWatcher.start();
-        } else {
-            configWatcher = null;
-        }
-
-        // CB-303 part 3: single ordered shutdown hook. Drain sessions first while herdr is still
-        // open (so releases reach the daemon), then stop poller/message/mcp/reaper, and close herdr
-        // last. This replaces the earlier independent hooks that could race and close herdr early.
-        Runtime.getRuntime().addShutdownHook(new Thread(() -> {
-            sessions.close(cfg.lifecycle() != null ? cfg.lifecycle().drainTimeoutSeconds() : null);
-            poller.stop();
-            messages.close();
-            pushLoop.close();
-            if (heartbeat != null) heartbeat.close(); // CB-551: stop the idle-lead heartbeat scheduler
-            if (healthMonitor != null) healthMonitor.stop();
-            if (configWatcher != null) configWatcher.stop(); // CB-559: stop polling the config file
-            mcp.close();
-            if (reaper != null) reaper.stop();
-            // Release the broker connection last among message resources (no-op for the in-memory inbox).
-            if (replyInbox instanceof AutoCloseable closeable) {
-                try {
-                    closeable.close();
-                } catch (Exception e) {
-                    log.debug("reply inbox close: {}", e.toString());
-                }
-            }
-            herdr.close();
-        }));
-
-        Javalin app = new BridgedApp(herdr, workers, sessions, messages, presence, mcp.servlet(),
-                callers, metrics).build();
-        app.start(cfg.bind().host(), cfg.bind().port());
-        log.info("bridged listening on {}:{}, herdr socket {}",
-                cfg.bind().host(), cfg.bind().port(), socket);
-    }
-
-    /**
-     * The {@link Injector}'s readiness gate (CB-534): a target is deliverable if it is a spawned
-     * member whose agent has connected the bridge MCP, <em>or</em> a lead.
-     *
-     * <p>The gate exists for one reason — to hold a delivery out of a <em>spawned</em> member's boot
-     * window, where herdr already reports {@code idle} but the TUI would drop an injected paste. That
-     * hazard is a property of spawning. A lead is never spawned: the operator started it and named it
-     * (or labelled its tab) only once it was up, so there is no boot window to guard.
-     *
-     * <p>A lead is also never enrolled in {@link MemberPresence} — {@code BridgeMcp} marks presence
-     * for every spawned member (worker and architect), deliberately, since that map doubles as the
-     * member roster's availability signal and a lead counted there would show up as an available
-     * member. So without the second disjunct a lead is permanently un-deliverable: every
-     * lead→lead send sat on the gate for {@code READINESS_GRACE_POLLS} (~60s) and then failed
-     * having never been typed into the pane.
-     *
-     * <p>The lead set is read through the supplier on each call rather than snapshotted, so a lead
-     * discovered by {@code leadScan} after startup becomes deliverable without a restart.
-     */
-    static Predicate<String> deliverableTo(MemberPresence presence, Supplier<Map<String, String>> leads) {
-        return target -> presence.isPresent(target) || leads.get().containsKey(target);
-    }
-
-    /**
-     * CB-594: which env vars the loaded config actually needs, and why — every non-{@code
-     * subscription} profile's {@code tokenEnv} (a subscription profile never reads one, see
-     * {@link BridgedConfig.Profile#isSubscription()}), plus every profile's {@code gitTokenEnv}
-     * where set (opt-in). Derived from the config, not hard-coded, so a new profile is covered for
-     * free. A var required by more than one profile is one entry naming every profile that needs
-     * it. Deliberately excludes {@code auth.tokenEnv}: that one is already enforced loudly, by a
-     * startup throw in {@code main()} — about 370 lines <em>below</em> this method's call site
-     * ({@link #reportRequiredSecrets(BridgedConfig)}), not a few lines above it. That throw only
-     * fires when {@code auth.mode: token} is configured; under the default loopback-trust mode it
-     * never runs, and {@code auth.tokenEnv} is simply not required.
-     *
-     * <p>Package-private and pure (no I/O, no logging) so the derivation is unit-testable without
-     * capturing log output; {@link #reportRequiredSecrets(BridgedConfig)} is the logging caller.
-     */
-    static Map<String, List<String>> requiredSecretEnvVars(BridgedConfig cfg) {
-        Map<String, List<String>> requiredBy = new LinkedHashMap<>();
-        cfg.profiles().forEach((name, profile) -> {
-            if (!profile.isSubscription()) {
-                requiredBy.computeIfAbsent(profile.tokenEnv(), _ -> new ArrayList<>())
-                        .add("profile '" + name + "' tokenEnv");
-            }
-            if (profile.hasGitToken()) {
-                requiredBy.computeIfAbsent(profile.gitTokenEnv(), _ -> new ArrayList<>())
-                        .add("profile '" + name + "' gitTokenEnv");
-            }
-        });
-        return requiredBy;
-    }
-
-    /**
-     * CB-594: log, by name only, which required env vars (see {@link #requiredSecretEnvVars}) are
-     * set in the daemon's own process environment — the environment every profile's {@code
-     * tokenEnv}/{@code gitTokenEnv} is read from at spawn time (see
-     * {@code HerdrPeerLauncher.resolveEnv}). Never logs a value, a prefix, or a length.
-     *
-     * <p>A missing entry only warns — it must never refuse to start. A daemon that boots and says
-     * what is wrong is strictly more useful than one that will not boot at all.
-     */
-    private static void reportRequiredSecrets(BridgedConfig cfg) {
-        Map<String, List<String>> requiredBy = requiredSecretEnvVars(cfg);
-        if (requiredBy.isEmpty()) {
-            log.info("startup secrets: no profile references a token env var — nothing to check");
-            return;
-        }
-        Map<String, String> env = System.getenv();
-        requiredBy.forEach((varName, sources) -> {
-            String value = env.get(varName);
-            if (value != null && !value.isBlank()) {
-                log.info("startup secret {}: set ({})", varName, String.join(", ", sources));
-            } else {
-                log.warn("startup secret {}: MISSING ({}) — the daemon will start anyway, and this "
-                                + "failure stays invisible until a worker actually needs it. Fix "
-                                + "${SHARED_ENV}/tools/secrets.sh and restart bridged from a LOGIN "
-                                + "shell (see scripts/redeploy-bridged.sh).",
-                        varName, String.join(", ", sources));
-            }
-        });
-    }
-
-    /**
-     * CB-596: {@code known:} empty (block absent entirely, or present but empty) means {@link
-     * BridgedConfig.MemberCredentials#blockedSet()} is empty too — every member pane inherits the
-     * operator's whole secret store, unblocked, exactly the defect this ticket fixes. Unlike a
-     * missing token ({@link #reportRequiredSecrets}), there is no name to point at: the point is
-     * that the block itself is missing. Warn once at startup and say what to add; never refuse to
-     * start over it — see {@link #reportRequiredSecrets} for why a daemon that boots and says
-     * what is wrong beats one that will not boot at all.
-     *
-     * <p>Package-private so the test can capture the log directly, the same way {@link
-     * #requiredSecretEnvVars} is exposed for {@link #reportRequiredSecrets}'s own test.
-     */
-    static void reportMemberCredentialsGap(BridgedConfig cfg) {
-        BridgedConfig.MemberCredentials creds = cfg.memberCredentials();
-        if (creds != null && !creds.known().isEmpty()) {
-            log.info("memberCredentials: {} known name(s), {} allowed — blocking {} on every spawn",
-                    creds.known().size(), creds.allow().size(), creds.blockedSet().size());
-            return;
-        }
-        log.warn("memberCredentials: absent or empty — the daemon will start anyway, and every "
-                + "member pane inherits the operator's WHOLE secret store, unblocked (CB-592's "
-                + "protection is lost). Add a memberCredentials: block (policy/allow/known) to "
-                + "bridged.yaml — see bridged.example.yaml — and restart.");
-    }
-
-    /**
-     * Poll herdr's {@code ping} until it answers or {@link #HERDR_WAIT_SECONDS} elapses (CB-504).
-     *
-     * @return true if herdr answered, false if it never did
-     */
-    private static boolean awaitHerdr(HerdrClient herdr) {
-        long deadline = System.nanoTime() + HERDR_WAIT_SECONDS * 1_000_000_000L;
-        boolean waited = false;
-        while (true) {
-            try {
-                herdr.call("ping");
-                if (waited) {
-                    log.info("herdr is up");
-                }
-                return true;
-            } catch (HerdrException e) {
-                if (System.nanoTime() >= deadline) {
-                    return false;
-                }
-                if (!waited) {
-                    log.info("waiting up to {}s for the herdr socket…", HERDR_WAIT_SECONDS);
-                    waited = true;
-                }
-                try {
-                    Thread.sleep(HERDR_WAIT_POLL_MILLIS);
-                } catch (InterruptedException ie) {
-                    Thread.currentThread().interrupt();
-                    return false;
-                }
-            }
-        }
-    }
-
-    private Bridged() {
-    }
-}

bridged/src/main/java/dev/ltms/bridged/auth/MemberLifecycle.java

@@ -1,21 +0,0 @@
-package dev.ltms.bridged.auth;
-
-import dev.ltms.bridged.peer.MemberRole;
-
-/** Optional session lifecycle hook for live member-slot bindings. */
-public interface MemberLifecycle {
-
-    MemberLifecycle NONE = new MemberLifecycle() {
-        @Override
-        public void acquired(MemberRole role, String profile, String terminal) {
-        }
-
-        @Override
-        public void released(String terminal) {
-        }
-    };
-
-    void acquired(MemberRole role, String profile, String terminal);
-
-    void released(String terminal);
-}

bridged/src/main/java/dev/ltms/bridged/herdr/PaneLocator.java

@@ -1,59 +0,0 @@
-package dev.ltms.bridged.herdr;
-
-import com.fasterxml.jackson.databind.JsonNode;
-
-import java.util.Map;
-
-/**
- * Resolves which herdr pane a process belongs to — the herdr half of connection-based MCP
- * identity (CB-105). Given the PID that opened an MCP connection, {@link #terminalForPid} finds
- * the agent pane whose process tree contains it, so {@code bridged} can tell <em>which worker</em>
- * is calling without the worker sending anything spoofable.
- *
- * <p>herdr owns the PID→pane truth: {@code pane.process_info} reports each pane's {@code shell_pid}
- * and foreground process PIDs. This scans agent panes; a spawn-time {@code pid→terminal} cache is
- * the obvious optimization once wired into {@code ClaudeCodeLauncher}.
- */
-public final class PaneLocator {
-
-    private final HerdrClient herdr;
-
-    public PaneLocator(HerdrClient herdr) {
-        this.herdr = herdr;
-    }
-
-    /**
-     * The {@code terminal_id} of the agent pane whose process tree contains {@code pid}, or
-     * {@code null} if no agent pane owns it (e.g. the caller is the primary, or off-host).
-     */
-    public String terminalForPid(long pid) {
-        if (pid <= 0) {
-            return null;
-        }
-        for (JsonNode pane : herdr.call("pane.list", Map.of()).path("panes")) {
-            String paneId = pane.path("pane_id").asText(null);
-            if (paneId != null && paneOwnsPid(paneId, pid)) {
-                return pane.path("terminal_id").asText(null);
-            }
-        }
-        return null;
-    }
-
-    private boolean paneOwnsPid(String paneId, long pid) {
-        JsonNode info;
-        try {
-            info = herdr.call("pane.process_info", Map.of("pane_id", paneId)).path("process_info");
-        } catch (HerdrException e) {
-            return false; // pane vanished mid-scan — just skip it
-        }
-        if (info.path("shell_pid").asLong(-1) == pid) {
-            return true;
-        }
-        for (JsonNode p : info.path("foreground_processes")) {
-            if (p.path("pid").asLong(-1) == pid) {
-                return true;
-            }
-        }
-        return false;
-    }
-}

bridged/src/main/java/dev/ltms/bridged/inject/CompletionResolver.java

@@ -1,372 +0,0 @@
-package dev.ltms.bridged.inject;
-
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.msg.Rendezvous;
-import dev.ltms.bridged.msg.TurnToken;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-
-import java.util.List;
-import java.util.Objects;
-import java.util.Set;
-import java.util.TreeSet;
-import java.util.concurrent.CompletableFuture;
-import java.util.concurrent.ConcurrentHashMap;
-import java.util.regex.Pattern;
-
-/**
- * The CB-106 completion fallback: bridges the {@link Injector}'s turn-completion signal to the
- * {@link Rendezvous} so a blocking {@code bridge_send} resolves even when the worker finishes its
- * task without ever calling {@code bridge_reply} — the common case for a real delegated coding task.
- *
- * <p>On a confirmed {@code working → idle} boundary it scrapes the worker's recent transcript and
- * resolves the awaiting send with that tail (a {@link Rendezvous.Kind#COMPLETION} resolution, so the
- * caller can tell a scrape from a structured reply). It scrapes only when a send is actually waiting
- * — a fleet worker's own turns, or a send that already timed out, cost no herdr traffic. An explicit
- * {@code bridge_reply} that raced in first wins; {@link Rendezvous#resolveCompletion} is then a no-op.
- *
- * <p>It also handles the CB-109 stall signal ({@link #onTurnFailed}): a worker that ran a turn then
- * wedged in an {@code unknown} state resolves the send as a failure (with the error screen as
- * context) rather than leaving it to time out.
- *
- * <p>The scrape is cleaned to the last {@code ⏺} assistant block (stripping TUI chrome) and guarded
- * against misattribution (CB-115): the pane content is baselined on delivery ({@link #onDelivered}),
- * and a completion whose scrape is unchanged from that baseline — the previous turn's wind-down
- * sampled as this turn's boundary on a rapid back-to-back send — is suppressed rather than resolving
- * the send with a stale answer.
- *
- * <p><strong>Waiter-specific resolution (CB-116).</strong> On delivery we also capture the exact
- * {@link Rendezvous} waiter this turn belongs to, and the completion/failure fallbacks resolve
- * <em>that</em> waiter — never "whatever send is waiting now". A completion fallback runs on a virtual
- * thread and can land after the worker's {@code bridge_reply} already resolved the turn and the
- * <em>next</em> send opened its own waiter on the same session; resolving the current waiter would
- * then deliver turn N's stale scrape as turn N+1's answer. Targeting the captured waiter makes a late
- * completion a harmless no-op (its waiter is already done) instead of a cross-turn stale reply.
- *
- * <p>Wired as the {@link Injector}'s {@link TurnListener}; the handlers hand off to a virtual thread
- * so the scrape's herdr round-trip never stalls the status poller. The captured waiter is read on the
- * poller thread (before any next-turn delivery can overwrite it) and passed into the virtual thread.
- */
-public final class CompletionResolver implements TurnListener {
-
-    private static final Logger log = LoggerFactory.getLogger(CompletionResolver.class);
-
-    /**
-     * herdr {@code agent.read} source for the completion scrape. {@code recent} returns the tail of
-     * the transcript (the worker's last output), which is what a delegator wants when the worker
-     * didn't structure a reply.
-     */
-    static final String SCRAPE_SOURCE = "recent";
-
-    /** Cap the scraped tail so a long transcript can't return an unbounded blob. */
-    static final int MAX_SCRAPE_CHARS = 4000;
-
-    private static final String CLIPPED_PANE_TAIL_MARKER =
-            "[Pane tail clipped: member did not call bridge_reply.]";
-
-    private final AgentControl agents;
-    private final Rendezvous rendezvous;
-    private final ExhaustedPatternLookup exhaustedPatterns;
-    private final ExhaustionSink exhaustionSink;
-
-    /**
-     * Per-target record of the turn currently in flight: the exact {@link Rendezvous} waiter its
-     * delivering send opened, plus the assistant block present when it was delivered.
-     *
-     * <p>The {@code waiter} is what makes a late fallback safe (CB-116): we resolve it, not "whoever
-     * is waiting now", so a completion that fires after the next send has opened its own waiter is a
-     * no-op rather than a cross-turn stale reply. The {@code baseline} is the CB-115 staleness
-     * reference: a completion scrape equal to it means the worker produced no new output (the previous
-     * turn's wind-down sampled as this boundary), so it is suppressed. Overwritten on each delivery;
-     * cleared when the turn resolves. Package-private so tests can capture and replay a specific turn.
-     */
-    record InFlight(CompletableFuture<Rendezvous.Resolution> waiter, String baseline) {
-    }
-
-    private final ConcurrentHashMap<String, InFlight> inFlight = new ConcurrentHashMap<>();
-
-    /**
-     * @param exhaustedPatterns CB-578 stage A: per-target lookup for a profile's configured
-     *                          usage-limit refusal pattern. Required — there is deliberately no
-     *                          defaulting overload; a caller that does not want the classification
-     *                          must pass an explicit inert value ({@link ExhaustedPatternLookup#none()}).
-     * @param exhaustionSink    CB-578 stage B: notified when a {@code BACKEND_EXHAUSTED}
-     *                          classification actually resolves a waiter. Required for the same
-     *                          reason as {@code exhaustedPatterns} — pass {@link ExhaustionSink#none()}
-     *                          to opt out.
-     */
-    public CompletionResolver(AgentControl agents, Rendezvous rendezvous, ExhaustedPatternLookup exhaustedPatterns,
-                              ExhaustionSink exhaustionSink) {
-        this.agents = agents;
-        this.rendezvous = rendezvous;
-        this.exhaustedPatterns = Objects.requireNonNull(exhaustedPatterns, "exhaustedPatterns");
-        this.exhaustionSink = Objects.requireNonNull(exhaustionSink, "exhaustionSink");
-    }
-
-    @Override
-    public void onDelivered(String target, TurnToken token) {
-        // Capture the exact waiter this turn belongs to (CB-116) and snapshot the pane's pre-turn
-        // content — what it shows *before* the just-delivered turn produces output — as the staleness
-        // reference (CB-115). Done synchronously (like the delivering send itself) so both are in
-        // place before this turn's completion can fire.
-        captureBaseline(target, token);
-    }
-
-    /** Capture the in-flight turn: its waiter and pre-turn baseline (the testable core of {@link #onDelivered}). */
-    void captureBaseline(String target, TurnToken token) {
-        CompletableFuture<Rendezvous.Resolution> waiter = token.waiter();
-        if (waiter == null) {
-            inFlight.remove(target); // no send is waiting on this delivery — nothing to resolve later
-            return;
-        }
-        String baseline;
-        try {
-            // Clip to the same cap resolve() applies to the tail (line ~134): the CB-115 misattribution
-            // guard compares baseline.equals(tail), so both sides must be the same capped representation.
-            // An unclipped baseline vs a clipped tail would never match for a >MAX_SCRAPE_CHARS block,
-            // defeating the guard and letting a stale completion resolve the send.
-            baseline = clip(lastAssistantBlock(agents.read(target, SCRAPE_SOURCE)));
-        } catch (RuntimeException e) {
-            baseline = null; // fail open: no baseline ⇒ no suppression
-            log.debug("delivery baseline for {} failed: {}", target, e.getMessage());
-        }
-        inFlight.put(target, new InFlight(waiter, baseline));
-    }
-
-    /** The turn currently baselined for {@code target}, or {@code null} — a test hook for the captureBaseline path. */
-    InFlight inFlight(String target) {
-        return inFlight.get(target);
-    }
-
-    @Override
-    public void onTurnComplete(String target) {
-        // Read the in-flight turn on the poller thread — before any next-turn delivery can overwrite
-        // it — then off-load the scrape (a herdr round-trip we must not block polling on) to a vthread.
-        InFlight turn = inFlight.get(target);
-        Thread.ofVirtual().name("completion-" + target).start(() -> resolve(target, turn));
-    }
-
-    /**
-     * Resolve the completed turn before adapter housekeeping can erase its rendered output. This is
-     * intentionally synchronous and used only when a post-turn context reset is enabled; the normal
-     * path remains off-loaded so polling is not blocked by a scrape.
-     */
-    public void resolveBeforePostAction(String target) {
-        resolve(target, inFlight.get(target));
-    }
-
-    @Override
-    public void onTurnFailed(String target) {
-        InFlight turn = inFlight.get(target);
-        Thread.ofVirtual().name("turn-failed-" + target).start(() -> fail(target, turn, null));
-    }
-
-    @Override
-    public void onTurnFailed(String target, String reason) {
-        InFlight turn = inFlight.get(target);
-        Thread.ofVirtual().name("turn-failed-" + target).start(() -> fail(target, turn, reason));
-    }
-
-    /** Synchronous resolve (the unit-testable core of {@link #onTurnComplete}). */
-    void resolve(String target, InFlight turn) {
-        CompletableFuture<Rendezvous.Resolution> waiter = turn == null ? null : turn.waiter();
-        if (waiter == null || waiter.isDone()) {
-            // Nobody is blocked on THIS turn (it had no send, or its bridge_reply already won). Skip
-            // the scrape; resolving the current waiter here would be the CB-116 cross-turn stale reply.
-            inFlight.remove(target, turn);
-            return;
-        }
-        String tail;
-        String assistantBlock = null;
-        int originalLength = 0;
-        boolean clipped = false;
-        boolean scrapeFailed = false;
-        try {
-            assistantBlock = lastAssistantBlock(agents.read(target, SCRAPE_SOURCE));
-            originalLength = assistantBlock.strip().length();
-            clipped = originalLength > MAX_SCRAPE_CHARS;
-            tail = clip(assistantBlock);
-        } catch (RuntimeException e) {
-            // The worker finished but we couldn't read its screen — still resolve the send so the
-            // caller unblocks; an empty tail beats hanging until the caller's timeout.
-            log.warn("completion scrape for {} failed; resolving with an empty tail: {}",
-                    target, e.getMessage());
-            tail = "";
-            scrapeFailed = true;
-        }
-        // Misattribution guard (CB-115): if the scrape is byte-identical to the pane content at
-        // delivery, this turn produced no new output — the boundary belongs to the previous turn's
-        // wind-down (common on rapid back-to-back sends). Suppress rather than resolve the send with
-        // a stale answer; the real bridge_reply (or a later genuine completion) resolves it instead.
-        // A scrape that failed to read is exempt — an empty tail there is "couldn't see", not "no change".
-        String baseline = turn.baseline();
-        if (!scrapeFailed && baseline != null && baseline.equals(tail)) {
-            log.debug("suppressing misattributed completion for {} (no output change since delivery)",
-                    target);
-            return; // keep the in-flight record: a later genuine completion still needs it
-        }
-        // CB-578 stage A: a turn that ended with no bridge_reply AND whose scrape matches the
-        // backend's configured usage-limit pattern is a refusal, not an answer. Classify it as
-        // BACKEND_EXHAUSTED rather than handing the caller a scrape that reads like a real reply.
-        if (!scrapeFailed) {
-            Pattern exhausted = exhaustedPatterns.patternFor(target);
-            String matchedLine = exhausted == null ? null : firstMatchingLine(assistantBlock, exhausted);
-            if (matchedLine != null) {
-                String reason = "backend exhausted (usage limit): " + matchedLine;
-                if (rendezvous.resolveExhausted(waiter, reason)) {
-                    inFlight.remove(target, turn);
-                    log.warn("completion for {} classified BACKEND_EXHAUSTED (no bridge_reply; scrape "
-                            + "matched the profile's exhausted pattern): {}", target, reason);
-                    // CB-578 stage B: only on the resolution that actually won the race — a late
-                    // duplicate must never quarantine a credential twice for one refusal.
-                    exhaustionSink.onExhausted(target, reason);
-                }
-                return;
-            }
-        }
-        String completion = clipped ? tail + "\n" + CLIPPED_PANE_TAIL_MARKER : tail;
-        if (rendezvous.resolveCompletion(waiter, completion)) {
-            inFlight.remove(target, turn);
-            if (clipped) {
-                log.warn("completion scrape for {} clipped from {} chars to the {} char cap; "
-                                + "member did not call bridge_reply, so the pane tail is partial",
-                        target, originalLength, MAX_SCRAPE_CHARS);
-            }
-            log.debug("resolved send to {} via turn-completion fallback ({} chars scraped)",
-                    target, tail.length());
-        }
-    }
-
-    /** Synchronous fail (the unit-testable core of {@link #onTurnFailed}). */
-    void fail(String target, InFlight turn) {
-        fail(target, turn, null);
-    }
-
-    /** Synchronous fail with an optional reason supplied by a dropped worker queue. */
-    void fail(String target, InFlight turn, String explicitReason) {
-        // A never-delivered readiness failure has no in-flight record but still has a blocked send;
-        // fall back to the currently-registered waiter (unambiguous — that send never completed, so
-        // no next turn exists to confuse it with).
-        CompletableFuture<Rendezvous.Resolution> waiter =
-                turn != null ? turn.waiter() : rendezvous.currentWaiter(target);
-        if (waiter == null || waiter.isDone()) {
-            inFlight.remove(target, turn); // nobody blocked on this worker — nothing to fail
-            return;
-        }
-        String reason = explicitReason;
-        if (reason == null || reason.isBlank()) {
-            try {
-                reason = clip(agents.read(target, SCRAPE_SOURCE));
-            } catch (RuntimeException e) {
-                reason = "";
-            }
-            if (reason.isBlank()) {
-                // No screen to scrape — either the worker is stuck (CB-109) or gone (CB-110).
-                reason = "worker did not reply; its turn ended in an unrecoverable state "
-                        + "(worker unreachable or stuck)";
-            }
-        }
-        if (rendezvous.resolveFailure(waiter, reason)) {
-            inFlight.remove(target, turn);
-            log.warn("failing send to {} via turn-stall fallback: {}", target, reason);
-        }
-    }
-
-    /**
-     * The first line of {@code text} matching {@code pattern}, stripped — the CB-578 stage A
-     * evidence carried in a {@code BACKEND_EXHAUSTED} reason so the operator sees the real refusal
-     * text, never a generic label. {@code null} if no line matches.
-     */
-    static String firstMatchingLine(String text, Pattern pattern) {
-        if (text == null || text.isEmpty()) return null;
-        for (String line : text.split("\n", -1)) {
-            if (pattern.matcher(line).find()) {
-                return line.strip();
-            }
-        }
-        return null;
-    }
-
-    /**
-     * Coverage summary for the CB-578 stage A exhausted-pattern classification, logged at startup
-     * the way {@link dev.ltms.bridged.health.FleetHealthMonitor#coverage} is — so an operator can
-     * see whether the classification is on, and for which profiles, without reading every
-     * profile's config by hand.
-     *
-     * @param allProfiles        every configured profile name
-     * @param configuredProfiles the subset of {@code allProfiles} that carry an exhausted pattern
-     */
-    public static String coverage(Set<String> allProfiles, Set<String> configuredProfiles) {
-        if (configuredProfiles.isEmpty()) {
-            return "off (no profile has an exhaustedPattern configured; profiles: " + sorted(allProfiles) + ")";
-        }
-        Set<String> unconfigured = new TreeSet<>(allProfiles);
-        unconfigured.removeAll(configuredProfiles);
-        return unconfigured.isEmpty()
-                ? "full (all profiles configured: " + sorted(allProfiles) + ")"
-                : "partial (configured: " + sorted(configuredProfiles) + "; not configured: " + sorted(unconfigured) + ")";
-    }
-
-    private static List<String> sorted(Set<String> names) {
-        return names.stream().sorted().toList();
-    }
-
-    private static String clip(String s) {
-        if (s == null) return "";
-        String trimmed = s.strip();
-        return trimmed.length() <= MAX_SCRAPE_CHARS
-                ? trimmed
-                : trimmed.substring(trimmed.length() - MAX_SCRAPE_CHARS);
-    }
-
-    /**
-     * Extract the last assistant message from a raw Claude Code pane scrape (CB-115). Claude Code
-     * prefixes each assistant turn with {@code ⏺}; the delegator wants that answer, not the TUI
-     * chrome around it. Take everything from the final {@code ⏺} onward and stop at the <em>first</em>
-     * hard interface boundary below it — the spinner/status line, input box, {@code ❯} prompt (which
-     * may echo the <em>next</em> turn's text), footer, or tips/warnings. Stopping at the first
-     * boundary (rather than trimming only trailing chrome) is what keeps a following turn's echoed
-     * prompt out of this reply. Blank lines are not boundaries, so a multi-paragraph answer survives;
-     * trailing blanks are trimmed at the end. With no {@code ⏺} marker (an unusual render) the whole
-     * text is scanned the same way, so we never lose the reply.
-     *
-     * <p>Package-private and pure so it is unit-testable without herdr.
-     */
-    static String lastAssistantBlock(String raw) {
-        if (raw == null || raw.isBlank()) return "";
-        int marker = raw.lastIndexOf('⏺');
-        String block = marker >= 0 ? raw.substring(marker + 1) : raw;
-        StringBuilder out = new StringBuilder();
-        int kept = 0;
-        for (String line : block.split("\n", -1)) {
-            if (isBoundary(line)) break; // first TUI boundary ends the assistant message
-            if (kept++ > 0) out.append('\n');
-            out.append(line);
-        }
-        return out.toString().strip();
-    }
-
-    /**
-     * A hard TUI boundary line that marks the end of an assistant message and the start of interface
-     * chrome (input box, prompt, spinner, footer, tips/warnings). Blank lines are <em>not</em>
-     * boundaries — an answer may contain them — so they are kept and trimmed only if trailing.
-     */
-    private static boolean isBoundary(String line) {
-        String t = line.strip();
-        if (t.isEmpty()) return false;
-        // A horizontal rule / all box-drawing separators (e.g. "──────").
-        if (t.chars().allMatch(c -> c == '─' || c == '—' || c == '━' || c == '═' || c == '-')) {
-            return true;
-        }
-        String lower = t.toLowerCase();
-        return t.startsWith("╭") || t.startsWith("│") || t.startsWith("╰") || t.startsWith("┌")
-                || t.startsWith("└") || t.startsWith("❯") || t.startsWith("⏵")
-                || t.startsWith("⎿") || t.startsWith("⚠")
-                // Status/spinner lines Claude Code renders below a settled or in-flight turn,
-                // e.g. "✻ Baked for 21s", "✶ Forming…".
-                || t.startsWith("✻") || t.startsWith("✳") || t.startsWith("✽") || t.startsWith("·")
-                || t.startsWith("●") || t.startsWith("◐") || t.startsWith("✢") || t.startsWith("✶")
-                || lower.contains("auto mode") || lower.contains("for shortcuts")
-                || lower.contains("esc to interrupt") || lower.contains("bypass permissions");
-    }
-}

bridged/src/main/java/dev/ltms/bridged/inject/ExhaustionSink.java

@@ -1,31 +0,0 @@
-package dev.ltms.bridged.inject;
-
-/**
- * Notified when {@link CompletionResolver} actually delivers a {@code BACKEND_EXHAUSTED}
- * classification to a waiting send (CB-578 stage B) — never on a race that lost (see
- * {@link CompletionResolver#resolve}, which only calls this after
- * {@code Rendezvous.resolveExhausted} returns {@code true}).
- *
- * <p>{@link CompletionResolver} knows only {@code target} (a herdr terminal id); it has no notion of
- * profiles or credentials, so mapping {@code target} to whatever should be quarantined is entirely
- * the sink's job — see {@code Bridged.main}'s wiring, which resolves target → session → profile →
- * {@code effectiveCredentialId()} and calls {@code BackendQuarantine.quarantine} on it.
- */
-@FunctionalInterface
-public interface ExhaustionSink {
-
-    /**
-     * @param target the herdr terminal id whose turn was classified {@code BACKEND_EXHAUSTED}
-     * @param reason the matched-line reason carried by the classification
-     */
-    void onExhausted(String target, String reason);
-
-    /**
-     * Inert sink — nothing happens on exhaustion. The explicit stand-in a caller (or a test not
-     * exercising this feature) passes instead of a defaulting overload, exactly like
-     * {@link ExhaustedPatternLookup#none()}.
-     */
-    static ExhaustionSink none() {
-        return (target, reason) -> { };
-    }
-}

bridged/src/main/java/dev/ltms/bridged/member/ClaudeCodeLauncher.java

@@ -1,373 +0,0 @@
-package dev.ltms.bridged.member;
-
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.guard.SubscriptionGuard;
-import dev.ltms.bridged.herdr.Agent;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.peer.Capability;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-
-import java.util.EnumSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.UUID;
-import java.util.function.Function;
-import java.util.function.LongSupplier;
-import java.util.function.Supplier;
-
-/**
- * The {@link HerdrPeerLauncher} adapter for <strong>Claude Code</strong> — the safe path from a
- * delegation request to a running off-subscription Claude.
- *
- * <p>Everything transport-related (tab/pane placement, the CB-306 spawn-readiness gate, unique
- * naming, CB-117 orphan reap, teardown, listing, cwd resolution) lives in the base. This class
- * supplies only the two Claude-specific seams:
- * <ul>
- *   <li>the {@code claude} name prefix (so reap matches {@code claude-*} panes, never another
- *       adapter's), and</li>
- *   <li>{@link #buildLaunch}, which encodes the subscription boundary: build the worker env with
- *       {@code ANTHROPIC_BASE_URL}, assert that host is on the allowlist <em>before</em> touching
- *       herdr, and mount the bridge MCP + reply charter as inline launch flags. A worker's base_url
- *       lives in the env map handed to herdr and nowhere else; {@code bridged}'s own environment is
- *       never mutated, and nothing is written to the worker's profile.</li>
- * </ul>
- */
-public final class ClaudeCodeLauncher extends HerdrPeerLauncher {
-
-    /** Label prefix for this adapter's herdr agent names (drives naming + orphan reap). */
-    private static final String NAME_PREFIX = "claude";
-
-    private static final Logger log = LoggerFactory.getLogger(ClaudeCodeLauncher.class);
-
-    private final SubscriptionGuard guard;
-
-    /**
-     * Production constructor — disables the spawn-ready gate ({@code spawnReadyTimeoutMs == 0}) so
-     * existing deployments and tests keep the legacy non-blocking spawn semantics.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env) {
-        this(agents, spaces, guard, profiles, defaultProfile, env, 0,
-                System::currentTimeMillis, () -> sleepUninterruptibly(300));
-    }
-
-    /**
-     * Production constructor with spawn-ready gate enabled. The gate polls {@code agents.status()}
-     * until the pane reports an injectable state or {@code spawnReadyTimeoutMs} elapses.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs, long spawnReadyPollMs) {
-        this(agents, spaces, guard, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, spawnReadyPollMs, null);
-    }
-
-    /**
-     * Production constructor carrying the fleet-wide tab-label template (CB-557). The template comes
-     * from {@code fleet.tabLabel}, which a profile cannot know because it names the member's
-     * <em>role</em>; a profile may still override it with its own {@code tabLabel}.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs, long spawnReadyPollMs,
-                              Supplier<BridgedConfig.Fleet> fleet) {
-        this(agents, spaces, guard, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs,
-                System::currentTimeMillis, () -> sleepUninterruptibly(spawnReadyPollMs),
-                fleet);
-    }
-
-    /**
-     * Production constructor, plus the CB-596 {@code memberCredentials} policy supplier.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs, long spawnReadyPollMs,
-                              Supplier<BridgedConfig.Fleet> fleet,
-                              Supplier<BridgedConfig.MemberCredentials> memberCredentials) {
-        this(agents, spaces, guard, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs,
-                System::currentTimeMillis, () -> sleepUninterruptibly(spawnReadyPollMs),
-                fleet, memberCredentials);
-    }
-
-    /**
-     * Full testability constructor. Every injectable collaborator is explicit so unit tests supply
-     * fakes for the clock ({@code nowMillis}) and poll-loop wait ({@code sleeper}). The
-     * {@code sleeper} is never called when the gate is disabled ({@code spawnReadyTimeoutMs == 0}).
-     *
-     * @param agents               herdr agent control (start, status, close)
-     * @param spaces               workspace / tab control (ensure, create, close)
-     * @param guard                subscription-boundary guard (checked before spawning)
-     * @param profiles             configured worker profiles
-     * @param defaultProfile       profile a no-argument spawn uses (nullable)
-     * @param env                  host env lookup (injectable for tests)
-     * @param spawnReadyTimeoutMs  max ms to wait for injectable state (0 disables the gate)
-     * @param nowMillis            monotonic clock source (e.g. {@code System::currentTimeMillis})
-     * @param sleeper              sleep/wait hook (e.g. {@code () -> Thread.sleep(pollMs)}); it
-     *                             already encodes the poll interval, so the 8th positional argument
-     *                             (poll ms) is accepted for API symmetry but otherwise unused here
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs,
-                              LongSupplier nowMillis, Runnable sleeper) {
-        this(agents, spaces, guard, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, null);
-    }
-
-    /**
-     * Full testability constructor, plus the fleet-wide tab-label template (CB-557).
-     *
-     * @param fleet live fleet config, read once for each spawn
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs,
-                              LongSupplier nowMillis, Runnable sleeper,
-                              Supplier<BridgedConfig.Fleet> fleet) {
-        super(NAME_PREFIX, agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, fleet);
-        this.guard = guard;
-    }
-
-    /**
-     * Full testability constructor, plus the CB-596 {@code memberCredentials} policy supplier.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs,
-                              LongSupplier nowMillis, Runnable sleeper,
-                              Supplier<BridgedConfig.Fleet> fleet,
-                              Supplier<BridgedConfig.MemberCredentials> memberCredentials) {
-        super(NAME_PREFIX, agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, fleet, memberCredentials);
-        this.guard = guard;
-    }
-
-    /**
-     * Full testability constructor, plus an injectable host-env-names source for the CB-596
-     * criterion-4 gap detector. Test seam only — every production call site leaves this at the
-     * default (the real {@code System.getenv()} key set) via the constructor above.
-     */
-    public ClaudeCodeLauncher(AgentControl agents, WorkspaceControl spaces, SubscriptionGuard guard,
-                              Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                              Function<String, String> env,
-                              long spawnReadyTimeoutMs,
-                              LongSupplier nowMillis, Runnable sleeper,
-                              Supplier<BridgedConfig.Fleet> fleet,
-                              Supplier<BridgedConfig.MemberCredentials> memberCredentials,
-                              Supplier<Set<String>> hostEnvNames) {
-        super(NAME_PREFIX, agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, fleet, memberCredentials, hostEnvNames);
-        this.guard = guard;
-    }
-
-    /**
-     * {@inheritDoc}
-     *
-     * <p>The spawn sequence encodes the subscription boundary: assert the profile's base_url is on
-     * the allowlist <em>before</em> any herdr call, then build the worker env with
-     * {@code ANTHROPIC_*}, the parity-neutral git-forge grant, and the bridge MCP + reply charter
-     * mounted as inline launch flags. When the request carries session identity (CB-547a) it is
-     * applied here — see {@link #applySessionIdentity}.
-     */
-    @Override
-    protected Launch buildLaunch(BridgedConfig.Profile cfg, LaunchSpec spec) {
-        // CB-539: a profile may deliberately opt into the subscription (subscription: true) when no
-        // off-subscription endpoint exists for it — e.g. `sonnet` on `ccs`. That profile gets no
-        // ANTHROPIC_BASE_URL/AUTH_TOKEN (there is nothing to point them at) and the guard's base_url
-        // requirement is skipped FOR IT ONLY. Every other profile keeps the hard boundary below.
-        boolean onSubscription = cfg.isSubscription();
-        String baseUrl = cfg.baseUrl();
-
-        if (onSubscription) {
-            // NO SILENT CONTRADICTION: subscription:true + a baseUrl state opposite intents; refuse
-            // loudly rather than pick a winner.
-            if (baseUrl != null && !baseUrl.isBlank()) {
-                throw new IllegalStateException("profile '" + cfg.profile()
-                        + "' sets both subscription: true and a baseUrl ('" + baseUrl + "') — the two "
-                        + "are contradictory: a subscription profile must not point at an endpoint. "
-                        + "Drop baseUrl, or drop subscription: true.");
-            }
-            // Visible without anyone going looking for it: this worker bills the subscription.
-            log.warn("spawning profile '{}' on the Claude subscription (subscription: true) — this "
-                    + "worker WILL bill the operator's subscription", cfg.profile());
-        } else {
-            guard.assertWorker(baseUrl); // hard stop before we spawn anything
-        }
-
-        Map<String, String> workerEnv = baseEnv(cfg);
-        if (onSubscription) {
-            // CB-542 belt-and-braces: on the subscription path no guard vets these two keys, and the
-            // profile's env: is layered in by baseEnv — so strip any that rode in there. Config load
-            // already rejects this (loudly, naming the profile); this makes the boundary hold even
-            // for a profile built in code that never passed through that validation.
-            workerEnv.remove("ANTHROPIC_BASE_URL");
-            workerEnv.remove("ANTHROPIC_AUTH_TOKEN");
-        } else {
-            workerEnv.put("ANTHROPIC_BASE_URL", baseUrl);
-            putIfPresent(workerEnv, "ANTHROPIC_AUTH_TOKEN", env.apply(cfg.tokenEnv()));
-        }
-        putIfPresent(workerEnv, "ANTHROPIC_MODEL", cfg.model());
-        putIfPresent(workerEnv, "CLAUDE_CONFIG_DIR", cfg.configDir());
-        applyGitToken(workerEnv, cfg);
-
-        // CB-547a: Claude Code can MINT its own session id, so bridged chooses it — a fresh spawn
-        // gets a UUID we pass as --session-id and return from agentSessionId(), so the resume
-        // handle is known BEFORE the agent has written anything; a resume spawn adopts its prior
-        // id via -r and passes no --session-id (the two conflict). Both are injected before the
-        // model flag so --model keeps outranking the operator's own argv.
-        // mutableArgv: argvWithBridge may hand back the profile's own (immutable) List.of when it
-        // has neither MCP nor a charter — session flags must be added into a list we own.
-        List<String> argv = mutableArgv(argvWithBridge(cfg, spec.charter()));
-        String agentSessionId = applySessionIdentity(argv, spec.sessionName(), spec.resumeSessionId());
-        return new Launch(workerEnv, argvWithModel(argv, cfg), agentSessionId);
-    }
-
-    /**
-     * Add the Claude-specific session-identity flags to {@code argv} and return the peer's OWN
-     * session id — the resume handle. A resume request passes the prior id via {@code -r} and
-     * returns that id; a fresh named session mints a new UUID, passes it via {@code --session-id},
-     * and returns the mint. The bridge's logical name rides along as {@code -n} when present. When
-     * <em>no</em> identity is requested (sessionName and resumeSessionId both blank) this adds
-     * nothing and returns {@code null}, keeping the legacy no-identity launch byte-identical.
-     */
-    private static String applySessionIdentity(List<String> argv, String sessionName, String resumeSessionId) {
-        boolean resuming = resumeSessionId != null && !resumeSessionId.isBlank();
-        boolean named = sessionName != null && !sessionName.isBlank();
-        if (!resuming && !named) {
-            return null; // no identity requested — keep the legacy launch byte-identical
-        }
-        if (named) {
-            argv.add("-n");
-            argv.add(sessionName);
-        }
-        if (resuming) {
-            argv.add("-r");
-            argv.add(resumeSessionId);
-            return resumeSessionId;
-        }
-        String minted = UUID.randomUUID().toString();
-        argv.add("--session-id");
-        argv.add(minted);
-        return minted;
-    }
-
-    /**
-     * The launch argv, plus an inline {@code --mcp-config} when {@code worker.mcpUrl} is set and
-     * {@code --append-system-prompt} when the base composed a charter. Neither touches the profile's
-     * config; both are pure command-line flags. This inline-flag mount is Claude Code specific —
-     * other adapters mount MCP and instructions their own way.
-     */
-    private List<String> argvWithBridge(BridgedConfig.Profile cfg, String charter) {
-        if (!cfg.hasMcp() && charter == null) {
-            return cfg.argv();
-        }
-        List<String> argv = mutableArgv(cfg.argv());
-        if (cfg.hasMcp()) {
-            String mcpJson = "{\"mcpServers\":{\"bridge\":{\"type\":\"http\",\"url\":\""
-                    + cfg.mcpUrl() + "\"}}}";
-            argv.add("--mcp-config");
-            argv.add(mcpJson);
-        }
-        if (charter != null) {
-            argv.add("--append-system-prompt");
-            argv.add(charter);
-        }
-        return argv;
-    }
-
-    /**
-     * Pin the model on the command line as well as in {@code ANTHROPIC_MODEL} (CB-533).
-     *
-     * <p>The env var alone is not a reliable pin for this adapter, because the argv is usually a
-     * launcher rather than {@code claude} itself — {@code ["ccs", "<profile>"]} — and {@code ccs}
-     * exports its profile's own model family ({@code ANTHROPIC_MODEL}, {@code DEFAULT_OPUS/SONNET/
-     * HAIKU}, {@code CLAUDE_CODE_SUBAGENT_MODEL}) over whatever it inherited. A worker profile that
-     * set {@code model:} therefore got silently overruled by its own launcher. Claude Code's
-     * {@code --model} flag outranks the environment, and {@code ccs <profile> [claude-args...]}
-     * passes trailing arguments through, so the flag survives the wrapper.
-     *
-     * <p>Appended last so it also outranks anything in the operator's own {@code argv}. Profiles
-     * that deliberately leave {@code model:} unset (letting {@code ccs} own model selection, as
-     * {@code gx10} does) are untouched — this adds nothing when there is nothing to add. This is
-     * the {@code kind: claude} counterpart of the opencode adapter's {@code -m provider/model}.
-     */
-    private static List<String> argvWithModel(List<String> argv, BridgedConfig.Profile cfg) {
-        if (cfg.model() == null || cfg.model().isBlank()) {
-            return argv;
-        }
-        List<String> withModel = mutableArgv(argv);
-        withModel.add("--model");
-        withModel.add(cfg.model());
-        return withModel;
-    }
-
-    // --- Agent-returning convenience spawns (used by callers/tests that want the herdr Agent) ---
-
-    /** Spawn a worker for the default profile in the resolved default cwd. */
-    public Agent spawn() {
-        return spawnInternal(null, null, null);
-    }
-
-    /** Spawn a worker for a named profile (null → default) in the resolved default cwd. */
-    public Agent spawn(String profileName) {
-        return spawnInternal(profileName, null, null);
-    }
-
-    /** Spawn a worker for a named profile with an explicit requested/caller cwd (CB-112). */
-    public Agent spawn(String profileName, String requestedCwd, String callerCwd) {
-        return spawnInternal(profileName, requestedCwd, callerCwd);
-    }
-
-    // --- capabilities --------------------------------------------------------------------------
-
-    @Override
-    public Set<Capability> capabilities() {
-        Set<Capability> caps = EnumSet.of(Capability.MID_TURN_ASK, Capability.WORKTREE,
-                Capability.CONTEXT_RESET, Capability.ORPHAN_REAP,
-                Capability.SESSION_NAME, Capability.SESSION_RESUME);
-        if (hasGitTokenProfile()) {
-            caps.add(Capability.SELF_PR);
-        }
-        return Set.copyOf(caps);
-    }
-
-    @Override
-    public boolean clearContext(String id) {
-        String target = agentTarget(id);
-        if (target == null) {
-            return false;
-        }
-        // This deliberately bypasses Injector: /clear is housekeeping, not a delegated turn.
-        agents().send(target, "/clear");
-        return true;
-    }
-
-    /** Whether any configured profile opts into a git-forge token (required for {@link Capability#SELF_PR}). */
-    private boolean hasGitTokenProfile() {
-        return profileConfigs().stream().anyMatch(BridgedConfig.Profile::hasGitToken);
-    }
-
-    // --- CB-117 reap predicate (Claude prefix), kept for direct unit testing -------------------
-
-    /**
-     * Whether {@code name} is a Claude Code bridge worker started by a <em>different</em> process
-     * than {@code currentNonce}. A thin {@code claude}-prefix binding of
-     * {@link HerdrPeerLauncher#isForeignWorker(String, String, String)}.
-     */
-    static boolean isForeignWorker(String name, String currentNonce) {
-        return HerdrPeerLauncher.isForeignWorker(NAME_PREFIX, name, currentNonce);
-    }
-}

bridged/src/main/java/dev/ltms/bridged/member/HerdrPeerLauncher.java

@@ -1,995 +0,0 @@
-package dev.ltms.bridged.member;
-
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.herdr.Agent;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.HerdrException;
-import dev.ltms.bridged.herdr.Tab;
-import dev.ltms.bridged.herdr.Workspace;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.peer.Capability;
-import dev.ltms.bridged.peer.CharterReceipt;
-import dev.ltms.bridged.peer.MemberRole;
-import dev.ltms.bridged.peer.PeerHandle;
-import dev.ltms.bridged.peer.PeerLauncher;
-import dev.ltms.bridged.peer.PeerUnreachableException;
-import dev.ltms.bridged.peer.SpawnRequest;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-
-import java.security.SecureRandom;
-import java.util.ArrayList;
-import java.util.Collection;
-import java.util.HashSet;
-import java.util.LinkedHashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.UUID;
-import java.util.concurrent.ConcurrentHashMap;
-import java.util.concurrent.ConcurrentMap;
-import java.util.concurrent.atomic.AtomicBoolean;
-import java.util.concurrent.atomic.AtomicLong;
-import java.util.function.Function;
-import java.util.function.LongSupplier;
-import java.util.function.Supplier;
-import java.util.regex.Matcher;
-import java.util.regex.Pattern;
-
-/**
- * Abstract base for {@link PeerLauncher} adapters that materialize a peer as a <em>herdr</em>
- * agent (a CLI coding agent running in a herdr tab/pane). It owns everything that is the same
- * regardless of <em>which</em> coding agent runs: tab/pane placement, the CB-306 spawn-readiness
- * gate, unique naming, CB-117 orphan reap, teardown, {@link #list() listing}, and cwd resolution.
- *
- * <p>Two seams are peer-specific and supplied by the concrete adapter:
- * <ul>
- *   <li>{@code namePrefix} (constructor arg) — the label prefix ({@code claude}, {@code opencode})
- *       that drives both unique naming and the orphan-reap pattern, so each adapter reaps only its
- *       own kind of pane and never another's.</li>
- *   <li>{@link #buildLaunch(BridgedConfig.Profile, LaunchSpec)} — the peer-specific env map + argv, including any
- *       subscription/guard check, MCP mount, and instruction injection. The base never sees how the
- *       peer is configured; it only places and starts the returned {@link Launch}.</li>
- * </ul>
- *
- * <p>Placement: in the default {@code tab} policy a peer lands in its own tab inside a dedicated
- * worker space (found-or-created once, then shared), so peers never split or clutter the user's
- * real work spaces. Teardown removes the peer's pane <em>and</em> its now-empty tab, tolerating an
- * already-gone peer so a repeated DELETE is harmless.
- */
-public abstract class HerdrPeerLauncher implements PeerLauncher {
-
-    private static final Logger log = LoggerFactory.getLogger(HerdrPeerLauncher.class);
-
-    /** herdr rejects a duplicate agent {@code name}; we retry a bumped name this many times. */
-    private static final int NAME_RETRIES = 8;
-
-    /**
-     * Retries for {@code agent.start} against a seed pane whose shell has not reached its prompt
-     * yet — {@code tab.create}/{@code pane.split} return as soon as the pane exists, and herdr
-     * refuses to start an agent in a pane that is not "an available shell" ({@code agent_pane_busy}).
-     */
-    private static final int SHELL_READY_RETRIES = 20;
-
-    private final String namePrefix;                       // label prefix: naming + reap scheme
-    private final AgentControl agents;
-    private final WorkspaceControl spaces;
-    private final Map<String, BridgedConfig.Profile> profiles; // profile name → spawn settings
-    private final String defaultProfile;                      // profile a no-arg spawn uses (nullable)
-
-    /** Host env lookup (injectable for tests); adapters read it in {@link #buildLaunch}. */
-    protected final Function<String, String> env;
-
-    private final AtomicLong nameSeq = new AtomicLong(); // per-peer counter (herdr agent names only)
-
-    /**
-     * Live fleet config, read once per spawn. A null supplier or value leaves tab labels at their
-     * default and supplies no role charter. A profile's own {@code tabLabel} still overrides it.
-     *
-     * <p>CB-559: a supplier rather than a snapshot, so a config reload affects the next launch
-     * without a restart. Existing tabs keep the label they were given.
-     */
-    private final Supplier<BridgedConfig.Fleet> fleet;
-
-    /**
-     * CB-596: the live {@code memberCredentials:} policy, read once per spawn (same hot-reload shape
-     * as {@link #fleet}). {@code null} — either the supplier itself, or what it returns — means no
-     * policy is configured and {@link #applyMemberCredentialPolicy} shadows nothing.
-     */
-    private final Supplier<BridgedConfig.MemberCredentials> memberCredentials;
-
-    /**
-     * Enumerates the daemon's own process environment variable NAMES ONLY, never values — the CB-596
-     * criterion-4 gap detector's data source (see {@link #logCredentialGap}). Injectable for tests;
-     * production always resolves to the real {@code System.getenv()} key set.
-     *
-     * <p>Deliberately the daemon's own environment, not the spawned pane's: nothing in the herdr
-     * client surface lets the daemon read back an arbitrary command's output from a pane before the
-     * peer starts in it, so there is no channel to inspect the pane's environment directly. The
-     * daemon's own process is started the same way (a login shell sourcing the same secret store —
-     * see CB-592's investigation of {@code secrets.sh}), so on a single-host deployment its env
-     * mirrors what the pane's login shell is about to export.
-     */
-    private final Supplier<Set<String>> hostEnvNames;
-
-    /** The final instruction always requires a bridge reply when the bridge MCP is mounted. */
-    protected static final String REPLY_CHARTER =
-            "You are a spawned member in the claude-bridge fleet. Every message you receive arrives "
-            + "through the bridge, and the ONLY channel back to the sender is the bridge_reply MCP tool. "
-            + "Text you write in your terminal is NOT sent anywhere — the sender cannot see your screen, "
-            + "so an in-terminal answer is silently discarded. Therefore you MUST end EVERY turn by calling "
-            + "bridge_reply with `content` set to your complete response. This holds for every message without "
-            + "exception — tasks, questions, clarifications, acknowledgements, and ordinary back-and-forth "
-            + "conversation. Call bridge_reply exactly once, as the final action of your turn, with your full "
-            + "answer in `content`; never wait for confirmation first. If you end a turn without calling "
-            + "bridge_reply, the sender receives nothing and the exchange stalls.";
-
-    /**
-     * Tab numbers, counted per {@code role/profile} pair (CB-557).
-     *
-     * <p>Deliberately not {@link #nameSeq}. That counter is shared by every profile this launcher
-     * serves, because its job is to make herdr <em>agent names</em> unique. Reusing it for the tab
-     * label made the numbers global, so sibling tabs read {@code #4}, {@code #9}, {@code #17} — gaps
-     * that look like a member died. Counting per role+profile makes {@code dev: sonnet #2} mean the
-     * second sonnet dev, which is what a reader assumes it means.
-     *
-     * <p>Resets when the daemon restarts, and that is fine: the label is a human-facing hint, not an
-     * identity. Identity is {@link PeerHandle#id()}.
-     */
-    private final ConcurrentMap<String, AtomicLong> labelSeq = new ConcurrentHashMap<>();
-
-    private final long spawnReadyTimeoutMs; // 0 = disable gate (legacy non-blocking spawn)
-    private final LongSupplier nowMillis; // monotonic clock (injectable for tests)
-    private final Runnable sleeper; // sleep/wait hook (injectable for tests; never real-sleep in unit tests)
-
-    // Per-process token mixed into each peer name so a fresh process (nameSeq back at 0) cannot
-    // collide with same-profile peers that outlived a restart. See startUniquelyNamed.
-    private final String nameNonce = String.format("%06x", new SecureRandom().nextInt(1 << 24));
-
-    // CB-519: PeerHandle.id() is a host-unique opaque UUID, decoupled from the herdr pane id. The
-    // routing/registry key is the UUID; the herdr pane id is a launcher-private placement/teardown
-    // coordinate. This map bridges the two so stop(id) can resolve a host-unique key back to the
-    // exact pane it must tear down. The pane id is launcher-private (never the routing key) — see
-    // PeerHandle.id().
-    private final ConcurrentMap<String, String> paneByAgentId = new ConcurrentHashMap<>();
-    private final AtomicBoolean resetUnsupportedLogged = new AtomicBoolean();
-
-    /**
-     * @param namePrefix          label prefix for this peer kind (drives naming and reap)
-     * @param agents              herdr agent control (start, status, close)
-     * @param spaces              workspace / tab control (ensure, create, close)
-     * @param profiles            configured peer profiles
-     * @param defaultProfile      profile a no-argument spawn uses (nullable)
-     * @param env                 host env lookup (injectable for tests)
-     * @param spawnReadyTimeoutMs max ms to wait for injectable state (0 disables the gate)
-     * @param nowMillis           monotonic clock source (e.g. {@code System::currentTimeMillis})
-     * @param sleeper             sleep/wait hook (never called when the gate is disabled); the poll
-     *                            interval is baked into this hook, so the base needs no poll field
-     */
-    protected HerdrPeerLauncher(String namePrefix, AgentControl agents, WorkspaceControl spaces,
-                                Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                                Function<String, String> env,
-                                long spawnReadyTimeoutMs,
-                                LongSupplier nowMillis, Runnable sleeper) {
-        this(namePrefix, agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                nowMillis, sleeper, null);
-    }
-
-    /**
-     * As above, plus the live {@code fleet} config (CB-557).
-     *
-     * @param fleet live fleet config, read once per spawn; {@code null} ⇒ default tab label and no
-     *              role charter. A separate constructor rather than a new parameter on the one
-     *              above, so every existing call site keeps the default without an edit.
-     */
-    protected HerdrPeerLauncher(String namePrefix, AgentControl agents, WorkspaceControl spaces,
-                                Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                                Function<String, String> env,
-                                long spawnReadyTimeoutMs,
-                                LongSupplier nowMillis, Runnable sleeper,
-                                Supplier<BridgedConfig.Fleet> fleet) {
-        this(namePrefix, agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                nowMillis, sleeper, fleet, null);
-    }
-
-    /**
-     * As above, plus the live {@code memberCredentials} policy (CB-596).
-     *
-     * @param memberCredentials live member-credential policy, read once per spawn; {@code null} ⇒
-     *                          no policy configured, so a spawn shadows nothing. A separate
-     *                          constructor rather than a new parameter on the one above, so every
-     *                          existing call site keeps the pre-CB-596 default without an edit.
-     */
-    protected HerdrPeerLauncher(String namePrefix, AgentControl agents, WorkspaceControl spaces,
-                                Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                                Function<String, String> env,
-                                long spawnReadyTimeoutMs,
-                                LongSupplier nowMillis, Runnable sleeper,
-                                Supplier<BridgedConfig.Fleet> fleet,
-                                Supplier<BridgedConfig.MemberCredentials> memberCredentials) {
-        this(namePrefix, agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                nowMillis, sleeper, fleet, memberCredentials, null);
-    }
-
-    /**
-     * As above, plus an injectable {@link #hostEnvNames} source for the CB-596 gap detector. Test
-     * seam only — every production call site leaves this {@code null} and gets the real host env.
-     */
-    protected HerdrPeerLauncher(String namePrefix, AgentControl agents, WorkspaceControl spaces,
-                                Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                                Function<String, String> env,
-                                long spawnReadyTimeoutMs,
-                                LongSupplier nowMillis, Runnable sleeper,
-                                Supplier<BridgedConfig.Fleet> fleet,
-                                Supplier<BridgedConfig.MemberCredentials> memberCredentials,
-                                Supplier<Set<String>> hostEnvNames) {
-        this.fleet = fleet;
-        this.namePrefix = namePrefix;
-        this.agents = agents;
-        this.spaces = spaces;
-        this.profiles = Map.copyOf(profiles);
-        this.defaultProfile = defaultProfile;
-        this.env = env;
-        this.spawnReadyTimeoutMs = spawnReadyTimeoutMs;
-        this.nowMillis = nowMillis;
-        this.sleeper = sleeper;
-        this.memberCredentials = memberCredentials;
-        this.hostEnvNames = hostEnvNames != null ? hostEnvNames : () -> System.getenv().keySet();
-    }
-
-    // --- adapter seams -------------------------------------------------------------------------
-
-    /**
-     * Build the peer-specific launch for {@code cfg}: the environment map and argv handed to herdr.
-     * Any subscription/guard check, MCP mount, and instruction injection happen here. The env map
-     * and argv are adapter-private; the base only places and starts what is returned.
-     */
-    protected abstract Launch buildLaunch(BridgedConfig.Profile cfg, LaunchSpec spec);
-
-    /** Direct transport access for peer-specific, non-turn control operations. */
-    protected final AgentControl agents() {
-        return agents;
-    }
-
-    /** Resolve the public peer id to the launcher's private herdr target. */
-    protected final String agentTarget(String id) {
-        return paneByAgentId.get(id);
-    }
-
-    @Override
-    public boolean clearContext(String id) {
-        if (resetUnsupportedLogged.compareAndSet(false, true)) {
-            log.warn("context reset is unsupported for peer kind {}; clearAfterTurn is a no-op",
-                    namePrefix);
-        }
-        return false;
-    }
-
-    /**
-     * A peer-specific launch: the herdr {@code env} map and {@code argv}, plus — for an adapter
-     * that carries durable session identity (CB-547a) — the peer's OWN session id
-     * ({@link PeerHandle#agentSessionId()}), known before the peer has written anything. Null for
-     * a launch that carries no identity.
-     */
-    protected record Launch(Map<String, String> env, List<String> argv, String agentSessionId) {
-
-        /** A launch without a discoverable agent session id (an adapter that carries none). */
-        Launch(Map<String, String> env, List<String> argv) {
-            this(env, argv, null);
-        }
-    }
-
-    /** All per-spawn values adapters may need, including the base-composed effective charter. */
-    protected record LaunchSpec(String sessionName, String resumeSessionId, MemberRole role, String charter) {
-    }
-
-    // --- profile surface -----------------------------------------------------------------------
-
-    /** The configured peer profile names (what {@code spawn(profile)} accepts). */
-    @Override
-    public Set<String> profiles() {
-        return profiles.keySet();
-    }
-
-    /** The parity-overlay file list for {@code profileName} (default list when unset). */
-    @Override
-    public List<String> parityOverlay(String profileName) {
-        String name = (profileName == null || profileName.isBlank()) ? defaultProfile : profileName;
-        if (name == null || name.isBlank()) {
-            return List.of();
-        }
-        BridgedConfig.Profile cfg = profiles.get(name);
-        return cfg == null ? List.of() : cfg.parityOverlay();
-    }
-
-    /** The profile a no-argument spawn uses, or {@code null} if none is configured. */
-    @Override
-    public String defaultProfile() {
-        return defaultProfile;
-    }
-
-    /**
-     * {@inheritDoc}
-     *
-     * <p>One {@link HerdrPeerLauncher} instance always serves exactly one adapter kind, so every
-     * profile it owns shares that adapter's {@link #capabilities()} — {@code profileName} only
-     * needs validating (throwing on an unknown profile, same as {@link #spawn}), not routing.
-     */
-    @Override
-    public Set<Capability> capabilitiesFor(String profileName) {
-        requireProfile(profileName);
-        return capabilities();
-    }
-
-    /** The configured profiles, for adapter capability decisions (e.g. any git-token grant). */
-    protected Collection<BridgedConfig.Profile> profileConfigs() {
-        return profiles.values();
-    }
-
-    /** Resolve {@code profileName} (null/blank → default) to its config, or throw with the options. */
-    protected BridgedConfig.Profile requireProfile(String profileName) {
-        String name = (profileName == null || profileName.isBlank()) ? defaultProfile : profileName;
-        if (name == null || name.isBlank()) {
-            throw new IllegalArgumentException("no default worker profile is configured — "
-                    + "pass a profile; configured: " + profiles.keySet());
-        }
-        BridgedConfig.Profile cfg = profiles.get(name);
-        if (cfg == null) {
-            throw new IllegalArgumentException("unknown worker profile '" + name
-                    + "' — configured: " + profiles.keySet());
-        }
-        return cfg;
-    }
-
-    // --- spawn ---------------------------------------------------------------------------------
-
-    /**
-     * A started peer plus the launch's agent-session id (the resume handle, or null) and the
-     * charter receipt (CB-571) the base composed for it.
-     */
-    private record Spawned(Agent agent, String agentSessionId, CharterReceipt receipt) {
-    }
-
-    /**
-     * Spawn a peer. {@code profileName} null/blank → the default profile. The working directory
-     * (CB-112) is resolved by {@link #resolveCwd}: an explicit {@code requestedCwd}, else the
-     * profile's configured {@code cwd}, else {@code callerCwd} (the primary's cwd, when the spawn
-     * came from the primary over MCP), else the daemon's cwd — never assumed to be {@code $HOME}.
-     * The adapter's {@link #buildLaunch} runs before any herdr call.
-     */
-    protected Agent spawnInternal(String profileName, String requestedCwd, String callerCwd) {
-        return spawnInternal(profileName, requestedCwd, callerCwd, null, null).agent();
-    }
-
-    /** Pre-CB-557 shape: no explicit role, so the tab is labelled as a {@code dev}. */
-    protected Spawned spawnInternal(String profileName, String requestedCwd, String callerCwd,
-                                    String sessionName, String resumeSessionId) {
-        return spawnInternal(profileName, requestedCwd, callerCwd, sessionName, resumeSessionId,
-                MemberRole.DEV);
-    }
-
-    /**
-     * Spawn a peer with session identity (CB-547a). The session values, role, and charter are
-     * threaded from the {@link SpawnRequest} into {@link #buildLaunch(BridgedConfig.Profile,
-     * LaunchSpec)}, and the launch's resolved agent-session id is returned alongside the agent so
-     * the caller can put it on the {@link PeerHandle}.
-     */
-    protected Spawned spawnInternal(String profileName, String requestedCwd, String callerCwd,
-                                    String sessionName, String resumeSessionId, MemberRole role) {
-        BridgedConfig.Profile cfg = requireProfile(profileName);
-        BridgedConfig.Fleet liveFleet = fleet == null ? null : fleet.get();
-        String roleCharter = liveFleet == null ? null : liveFleet.charterFor(role);
-        String replyCharter = cfg.hasMcp() ? REPLY_CHARTER : null;
-        String charter = roleCharter == null ? replyCharter
-                : replyCharter == null ? roleCharter : roleCharter + "\n\n" + replyCharter;
-        // CB-571: fingerprint the exact composed charter bytes once, here in the base, before the
-        // string leaves for an adapter — so Claude and OpenCode derive the same digest. A failed
-        // start has no bridge_spawn result and no roster row, so the failure log below is the only
-        // surface the byte count can appear on. The charter text itself is never logged.
-        CharterReceipt receipt = CharterReceipt.compose(role, cfg.profile(), roleCharter, charter);
-        try {
-            Launch launch = buildLaunch(cfg, new LaunchSpec(sessionName, resumeSessionId, role, charter));
-            String cwd = resolveCwd(requestedCwd, cfg, callerCwd);
-            Agent agent = cfg.tabPlacement()
-                    ? spawnInTab(cfg, launch.env(), launch.argv(), cwd, role, liveFleet)
-                    : spawnAsPane(cfg, launch.env(), launch.argv(), cwd, charter);
-            logCharterReceipt(receipt, true);
-            return new Spawned(agent, launch.agentSessionId(), receipt);
-        } catch (RuntimeException e) {
-            logCharterReceipt(receipt, false);
-            throw e;
-        }
-    }
-
-    /**
-     * The one place the charter's size and digest appear in the logs. {@code success} true after a
-     * start, false from the failure path of {@link #spawnInternal} where no handle or roster row
-     * exists to carry the receipt. Always metadata only — never the charter text.
-     */
-    private static void logCharterReceipt(CharterReceipt receipt, boolean success) {
-        String role = receipt.role() == null ? "" : receipt.role().wireName();
-        if (success) {
-            log.info("spawned role={} profile={} charterSource={} charterSha256={} charterBytes={}",
-                    role, receipt.profile(), receipt.charterSource(),
-                    receipt.charterSha256(), receipt.charterBytes());
-        } else {
-            log.warn("spawn failed; charter role={} profile={} charterSource={} charterSha256={} charterBytes={}",
-                    role, receipt.profile(), receipt.charterSource(),
-                    receipt.charterSha256(), receipt.charterBytes());
-        }
-    }
-
-    /**
-     * The next tab number for {@code role} on {@code profile}, starting at 1.
-     *
-     * <p>Starts at 1 rather than 0 because the number is read by a person: {@code "dev: sonnet #1"}
-     * is the first one, and {@code #0} invites the question of where {@code #1} went.
-     */
-    private long nextLabelSeq(MemberRole role, String profile) {
-        String key = (role == null ? "" : role.wireName()) + "/" + profile;
-        return labelSeq.computeIfAbsent(key, _ -> new AtomicLong()).incrementAndGet();
-    }
-
-    /**
-     * {@inheritDoc}
-     *
-     * <p>Delegates to {@link #spawnInternal} and wraps the resulting herdr {@link Agent} in a
-     * {@link WorkerHandle} whose {@link PeerHandle#id()} is a fresh <em>host-unique</em> opaque
-     * UUID (CB-519), deliberately decoupled from the herdr pane id: the id is the registry/routing
-     * key and must never collide across daemon processes on the same host, while the herdr pane id
-     * stays a launcher-private placement/teardown coordinate, remembered here so {@link #stop}
-     * can resolve the host-unique key back to its pane. When {@code spawnReadyTimeoutMs > 0},
-     * blocks until the peer's herdr status is injectable or the timeout elapses; on timeout the
-     * pane is closed (no orphan) and a {@link PeerUnreachableException} is thrown.
-     */
-    @Override
-    public PeerHandle spawn(SpawnRequest req) {
-        Spawned spawned = spawnInternal(req.profileName(), req.requestedCwd(), req.callerCwd(),
-                req.sessionName(), req.resumeSessionId(), req.role());
-        Agent agent = spawned.agent();
-        String paneId = agent.paneId();
-        if (spawnReadyTimeoutMs > 0) {
-            waitUntilInjectableOrThrow(paneId);
-        }
-        // CB-519: the handle id is a host-unique UUID; the herdr pane it maps to stays internal.
-        String id = UUID.randomUUID().toString();
-        paneByAgentId.put(id, paneId);
-        return new WorkerHandle(id, agent.terminalId(), requireProfile(req.profileName()).profile(),
-                req.sessionName(), spawned.agentSessionId(), spawned.receipt());
-    }
-
-    @Override
-    public String effectiveCwd(SpawnRequest req) {
-        return effectiveCwd(req.profileName(), req.requestedCwd(), req.callerCwd());
-    }
-
-    /**
-     * CB-301: the effective working directory a spawn for {@code profileName} would use, without
-     * actually spawning.
-     */
-    private String effectiveCwd(String profileName, String requestedCwd, String callerCwd) {
-        return resolveCwd(requestedCwd, requireProfile(profileName), callerCwd);
-    }
-
-    /**
-     * CB-112 cwd resolution: spawn arg → profile config → the primary's cwd → the daemon's cwd.
-     * Never returns {@code null}/blank: {@code "."} (the daemon's own working directory) is the
-     * guaranteed last resort so a pathological environment with an unset {@code user.dir} still
-     * honours the "never assume {@code $HOME}" contract rather than letting herdr default the pane.
-     */
-    private static String resolveCwd(String requestedCwd, BridgedConfig.Profile cfg, String callerCwd) {
-        return firstNonBlank(requestedCwd, cfg.cwd(), callerCwd, System.getProperty("user.dir"), ".");
-    }
-
-    private static String firstNonBlank(String... values) {
-        for (String v : values) {
-            if (v != null && !v.isBlank()) return v;
-        }
-        return null;
-    }
-
-    /** Dedicated worker space → own tab (carrying cwd+env) → start the peer into the seed pane. */
-    private Agent spawnInTab(BridgedConfig.Profile cfg, Map<String, String> workerEnv,
-                             List<String> argv, String cwd, MemberRole role, BridgedConfig.Fleet liveFleet) {
-        Workspace space = spaces.ensureWorkspace(cfg.workspace());
-        Tab.Created tab = spaces.createTab(space.workspaceId(), cwd, workerEnv);
-        log.info("spawning {} profile={} space={} tab={} cwd={}",
-                namePrefix, cfg.profile(), space.workspaceId(), tab.tab().tabId(), cwd);
-
-        Started started;
-        try {
-            if (tab.rootPaneId() == null) {
-                // Protocol 19 starts the agent INTO the seed pane — without one there is nowhere
-                // to start, and a partial tab would be left behind.
-                throw new IllegalStateException("tab " + tab.tab().tabId()
-                        + " had no seed pane in the create response — cannot start a peer in it");
-            }
-            started = startUniquelyNamed(cfg, argv, tab.rootPaneId());
-        } catch (RuntimeException e) {
-            // The peer never started — don't leave the tab we just created orphaned.
-            // Best-effort cleanup; never let it mask the real spawn failure.
-            try {
-                spaces.closeTab(tab.tab().tabId());
-            } catch (RuntimeException cleanup) {
-                log.warn("failed to close orphaned tab {} after spawn error: {}",
-                        tab.tab().tabId(), cleanup.getMessage());
-            }
-            throw e;
-        }
-
-        // The peer is LIVE now, in the seed pane itself (no shell pane to drop — protocol 19).
-        // Labelling is cosmetic: it must not fail the spawn or orphan the running peer — on error
-        // we log and still return it so the caller gets its paneId and can tear it down.
-        tidy("label tab " + tab.tab().tabId(),
-                () -> spaces.renameTab(tab.tab().tabId(),
-                        cfg.renderTabLabel(
-                                liveFleet == null ? null : liveFleet.tabLabel(),
-                                role, nextLabelSeq(role, cfg.profile()))));
-        log.info("{} started pane={} tab={} terminal={}",
-                namePrefix, started.agent().paneId(), started.agent().tabId(), started.agent().terminalId());
-        return started.agent();
-    }
-
-    /** Run a best-effort post-start cleanup step, logging (not throwing) on failure. */
-    private void tidy(String what, Runnable step) {
-        try {
-            step.run();
-        } catch (RuntimeException e) {
-            log.warn("post-start step failed ({}) — peer is running regardless: {}", what, e.getMessage());
-        }
-    }
-
-    /**
-     * Legacy placement: split the currently-focused tab; the peer still starts in {@code cwd}.
-     *
-     * <p>CB-571: this is the one legacy log that printed the full argv, and the charter travels
-     * inside argv — so the charter text went to the daemon log on every pane-placement spawn. The
-     * {@code spawnInTab} path never logs argv, so only this site is fixed. {@code charter} is the
-     * composed charter, if any; its argv element is replaced by its digest so the log still shows
-     * which args were passed without exposing the charter prose.
-     */
-    private Agent spawnAsPane(BridgedConfig.Profile cfg, Map<String, String> workerEnv,
-                              List<String> argv, String cwd, String charter) {
-        log.info("spawning {} (pane placement) profile={} cwd={} argv={}",
-                namePrefix, cfg.profile(), cwd, redactCharter(argv, charter));
-        String paneId = spaces.splitPane(cwd, workerEnv);
-        if (paneId == null) {
-            throw new IllegalStateException("pane.split returned no pane — cannot start a peer");
-        }
-        Agent peer = startUniquelyNamed(cfg, argv, paneId).agent();
-        log.info("{} started pane={} terminal={}", namePrefix, peer.paneId(), peer.terminalId());
-        return peer;
-    }
-
-    /**
-     * A copy of {@code argv} with an element equal to {@code charter} replaced by its digest, so
-     * the pane log never prints the charter prose. The charter is handed to an adapter as one argv
-     * element, so exact-equality is the right match; every other argument passes through unchanged.
-     */
-    private static List<String> redactCharter(List<String> argv, String charter) {
-        if (charter == null || charter.isBlank() || argv == null || argv.isEmpty()) {
-            return argv;
-        }
-        String digest = CharterReceipt.digestOf(charter);
-        return argv.stream()
-                .map(a -> a.equals(charter) ? "<charter sha256=" + digest + ">" : a)
-                .toList();
-    }
-
-    /** A started peer together with the sequence its unique name/label used. */
-    private record Started(Agent agent, long seq) {
-    }
-
-    /**
-     * Start the peer under a unique herdr agent name. herdr requires each running agent's
-     * {@code name} to be distinct (a 2nd identical {@code name} fails {@code agent_name_taken}) —
-     * the exact case that makes multiple peers useful. The name is
-     * {@code <prefix>-<profile>-<nonce>-<seq>}: {@code seq} distinguishes peers within this process,
-     * and the per-process {@code nonce} keeps a fresh process (whose {@code seq} restarts at 0) from
-     * colliding with same-profile peers that outlived a restart. The retry is a belt-and-braces
-     * backstop for the astronomically unlikely nonce+seq clash; the name is a label only — herdr
-     * detects kind and status from terminal output, not from it.
-     */
-    private Started startUniquelyNamed(BridgedConfig.Profile cfg, List<String> argv, String paneId) {
-        // Protocol 19 resolves the executable from the agent kind (== namePrefix here), so
-        // argv[0] — the configured executable — is dropped and only the extra args are passed.
-        List<String> args = argv.isEmpty() ? argv : argv.subList(1, argv.size());
-        HerdrException last = null;
-        for (int attempt = 0; attempt < NAME_RETRIES; attempt++) {
-            long seq = nameSeq.incrementAndGet();
-            String name = namePrefix + "-" + cfg.profile() + "-" + nameNonce + "-" + seq;
-            try {
-                return new Started(startAwaitingShellPrompt(name, args, paneId), seq);
-            } catch (HerdrException e) {
-                if (!"agent_name_taken".equals(e.code())) throw e;
-                log.debug("peer name '{}' taken, retrying", name);
-                last = e;
-            }
-        }
-        throw last;
-    }
-
-    /** Start the agent into {@code paneId}, waiting out the seed shell's boot with the sleeper. */
-    private Agent startAwaitingShellPrompt(String name, List<String> args, String paneId) {
-        HerdrException busy = null;
-        for (int attempt = 0; attempt < SHELL_READY_RETRIES; attempt++) {
-            try {
-                return agents.start(name, namePrefix, args, paneId);
-            } catch (HerdrException e) {
-                if (!"agent_pane_busy".equals(e.code())) throw e;
-                log.debug("pane {} not at its shell prompt yet, retrying agent.start", paneId);
-                busy = e;
-                sleeper.run();
-            }
-        }
-        throw busy;
-    }
-
-    // --- discovery + reap ----------------------------------------------------------------------
-
-    /** All herdr-tracked agents — discovery for "what peers exist". */
-    @Override
-    public List<Agent> list() {
-        return agents.list();
-    }
-
-    /**
-     * Reap peer panes left behind by an earlier daemon process (CB-117). herdr keeps a peer's pane
-     * alive across a daemon restart <em>by design</em>, and that pane's id is held only by its
-     * spawner — so a peer whose owning process exited before issuing the matching teardown leaks
-     * with nothing tracking it. On boot we scan herdr for agents whose name matches our
-     * {@code <prefix>-<profile>-<nonce>-<seq>} scheme with a nonce <em>other</em> than this
-     * process's {@link #nameNonce}, and tear each one down (its pane and, via {@link #stop}, its
-     * now-empty dedicated tab). A current-nonce peer is ours and live, so it is left running; a
-     * user's own session carries no such name and is never touched. A peer from a <em>different</em>
-     * adapter (different prefix) is likewise never touched. Best-effort: a failed listing, or a
-     * failure to stop any one peer, is logged and never aborts startup.
-     *
-     * @return the number of orphaned peers reaped
-     */
-    @Override
-    public int reapOrphanWorkers() {
-        List<Agent> all;
-        try {
-            all = agents.list();
-        } catch (RuntimeException e) {
-            log.warn("orphan-peer reap skipped — agent.list failed: {}", e.getMessage());
-            return 0;
-        }
-        int reaped = 0;
-        for (Agent a : all) {
-            if (!isForeignWorker(namePrefix, a.name(), nameNonce)) continue;
-            try {
-                stop(a.paneId());
-                reaped++;
-                log.info("reaped orphan {} {} (pane={} tab={}) left by a prior daemon",
-                        namePrefix, a.name(), a.paneId(), a.tabId());
-            } catch (RuntimeException e) {
-                log.warn("could not reap orphan {} {} (pane={}): {}",
-                        namePrefix, a.name(), a.paneId(), e.getMessage());
-            }
-        }
-        if (reaped > 0) {
-            log.info("orphan-peer reap complete — {} stale {} peer(s) removed at startup", reaped, namePrefix);
-        }
-        return reaped;
-    }
-
-    /** The {@code <prefix>-<profile>-<nonce>-<seq>} name pattern; group 1 captures the 6-hex nonce. */
-    static Pattern workerNamePattern(String prefix) {
-        return Pattern.compile(prefix + "-.*-([0-9a-f]{6})-\\d+");
-    }
-
-    /**
-     * Whether {@code name} is a peer of kind {@code prefix} started by a <em>different</em> process
-     * than {@code currentNonce} — the reap predicate (CB-117). True only for the prefix's naming
-     * scheme with a foreign nonce: a non-peer name, a different adapter's name, or our own live
-     * nonce is excluded. Pure and package-private so the decision is unit-testable without herdr.
-     */
-    static boolean isForeignWorker(String prefix, String name, String currentNonce) {
-        String nonce = workerNonce(prefix, name);
-        return nonce != null && !nonce.equals(currentNonce);
-    }
-
-    /** The 6-hex nonce embedded in a {@code prefix} peer name, or {@code null} if not one. */
-    static String workerNonce(String prefix, String name) {
-        if (name == null) return null;
-        Matcher m = workerNamePattern(prefix).matcher(name);
-        return m.matches() ? m.group(1) : null;
-    }
-
-    /** This process's peer-name nonce (a label component only; exposed for reaper tests). */
-    String nameNonce() {
-        return nameNonce;
-    }
-
-    // --- teardown ------------------------------------------------------------------------------
-
-    /**
-     * Tear a peer down: close the pane, and close its tab <em>only</em> when the peer is that tab's
-     * sole occupant. The single-pane check is what makes this safe regardless of how the peer was
-     * placed (or a placement-config change across a restart): a pane-placement peer sitting in one
-     * of the user's shared tabs has siblings, so its tab is never closed — we only ever remove a
-     * tab we created to hold one peer.
-     *
-     * <p>{@code idOrPane} is the {@link PeerHandle#id()} of a peer this launcher spawned (CB-519's
-     * host-unique opaque UUID), resolved through {@link #paneByAgentId} to the pane it must tear
-     * down. An argument that is not one of our ids is treated as a raw herdr pane id — the
-     * {@link #reapOrphanWorkers() orphan-reap} and spawn-gate-timeout paths, plus any caller that
-     * passes a pane directly, keep working without an owning id.
-     *
-     * <p>Resolves the tab from the pane <em>before</em> closing it. An already-gone pane/tab
-     * (repeated DELETE, crashed peer) is treated as success; any other failure propagates so a
-     * genuinely failed teardown is not reported as done.
-     */
-    @Override
-    public void stop(String idOrPane) {
-        // Teardown knows only the pane, not which profile spawned it. Attempt tab cleanup when any
-        // profile uses tab placement (so the bridge may have created a dedicated peer tab); the
-        // single-occupant check below is what actually protects the user's shared tabs.
-        String paneId = paneByAgentId.remove(idOrPane);
-        if (paneId == null) {
-            paneId = idOrPane; // raw-pane fallback (reap, gate timeout, pane-addressed callers)
-        }
-        WorkspaceControl.PaneLocation loc = usesTabPlacement() ? spaces.locatePane(paneId) : null;
-        try {
-            agents.close(paneId);
-        } catch (HerdrException e) {
-            if (!isAlreadyGone(e)) throw e;
-            log.debug("pane.close({}) ignored — already gone: {}", paneId, e.getMessage());
-        }
-        if (loc != null && loc.tabPaneCount() == 1) {
-            spaces.closeTab(loc.tabId());
-        } else if (loc != null) {
-            log.debug("not closing tab {} — it holds {} panes (not a dedicated peer tab)",
-                    loc.tabId(), loc.tabPaneCount());
-        }
-    }
-
-    /** Whether any configured profile places peers in their own tab (so tabs may need cleanup). */
-    private boolean usesTabPlacement() {
-        return profiles.values().stream().anyMatch(BridgedConfig.Profile::tabPlacement);
-    }
-
-    /** True when a herdr error means the target is already gone (safe to treat as done). */
-    private static boolean isAlreadyGone(HerdrException e) {
-        return e.code() != null && e.code().endsWith("_not_found");
-    }
-
-    // --- spawn-readiness gate (CB-306) ---------------------------------------------------------
-
-    /**
-     * Poll {@link AgentControl#status} until the pane reports an injectable state or the configured
-     * timeout elapses. On timeout, close the pane (self-reap) and throw.
-     */
-    private void waitUntilInjectableOrThrow(String paneId) {
-        long deadline = nowMillis.getAsLong() + spawnReadyTimeoutMs;
-        while (nowMillis.getAsLong() < deadline) {
-            if (agents.status(paneId).injectable()) {
-                log.debug("peer pane={} reached injectable state", paneId);
-                return;
-            }
-            sleeper.run();
-        }
-        log.warn("peer pane={} did not become injectable within {}ms — closing", paneId, spawnReadyTimeoutMs);
-        stop(paneId);
-        throw new PeerUnreachableException(
-                "worker pane " + paneId + " did not reach injectable state within "
-                        + spawnReadyTimeoutMs + "ms");
-    }
-
-    /**
-     * A concrete {@link PeerHandle} wrapping herdr agent coordinates, the profile that spawned it,
-     * the session identity the launch resolved (CB-547a): the bridge's logical name and the peer's
-     * own session id, both null when the spawn carried no identity — and the charter receipt
-     * (CB-571) the base computed for this launch.
-     */
-    private record WorkerHandle(String id, String terminalId, String profile,
-                                String sessionName, String agentSessionId,
-                                CharterReceipt receipt) implements PeerHandle {
-
-        @Override
-        public CharterReceipt charterReceipt() {
-            return receipt;
-        }
-    }
-
-    // --- shared helpers ------------------------------------------------------------------------
-
-    /** Put {@code k → v} only when {@code v} is present (non-null, non-blank). */
-    protected static void putIfPresent(Map<String, String> m, String k, String v) {
-        if (v != null && !v.isBlank()) {
-            m.put(k, v);
-        }
-    }
-
-    /** Host env lookup that tolerates an unconfigured (null/blank) var name — returns null then. */
-    protected String resolveEnv(String name) {
-        return (name == null || name.isBlank()) ? null : env.apply(name);
-    }
-
-    /**
-     * The parity-neutral git-forge token grant (CB-302): when {@code cfg} opts in via
-     * {@code gitTokenEnv} and the token resolves, inject {@code GITEA_TOKEN} plus its paired
-     * {@code GITEA_HOST}. Push over SSH is unaffected; the only incremental grant is PR-create.
-     * Peer-neutral, so every herdr adapter reuses it unchanged.
-     */
-    protected void applyGitToken(Map<String, String> workerEnv, BridgedConfig.Profile cfg) {
-        if (!cfg.hasGitToken()) {
-            return;
-        }
-        String gitToken = resolveEnv(cfg.gitTokenEnv());
-        if (gitToken != null) {
-            workerEnv.put("GITEA_TOKEN", gitToken);
-            putIfPresent(workerEnv, "GITEA_HOST", resolveEnv(cfg.gitHostEnv()));
-        }
-    }
-
-    /**
-     * CB-596: overlay value that shadows any host credential a herdr pane otherwise inherits from
-     * herdr's own login-shell process environment (gitea issue #82, superseding CB-592's single
-     * hardcoded {@code GITEA_ACCESS_TOKEN} name — see {@link #applyMemberCredentialPolicy}). herdr
-     * spawns a pane from its <em>own</em> process environment and layers our map on top —
-     * {@link dev.ltms.bridged.herdr.WorkspaceControl#createTab} and {@code #splitPane} send only
-     * the keys we put in that map, so any key we never mention passes straight through from
-     * herdr's own shell, admin credentials included.
-     *
-     * <p>Deliberately a non-blank sentinel, not {@code ""}. Whether an empty-string overlay value
-     * overrides an inherited variable or is skipped as blank could not be settled by reading this
-     * codebase — herdr's server-side merge is an external process, not something in this repo.
-     * A non-blank replacement sidesteps that ambiguity entirely: {@link #baseEnv}'s own {@code
-     * PATH} seeding already depends on the overlay reliably replacing an inherited value (see its
-     * javadoc), and that is only demonstrated for a non-blank value, so this reuses the same,
-     * proven-reliable shape rather than the unverified one.
-     *
-     * <p><b>MEASURED ON A LIVE PANE, 2026-08-15 (CB-592): this sentinel alone does NOT hold.</b> The
-     * overlay itself works — {@code GITEA_TOKEN} is injected here, is exported by no shell file, and
-     * does reach the pane. The sentinel loses one step later. A herdr pane runs a <em>login</em>
-     * shell, {@code ~/.zprofile} sources {@code ${SHARED_ENV}/tools/secrets.sh}, and that file does a
-     * plain unconditional {@code export GITEA_ACCESS_TOKEN=...}. A login shell overwrites a value
-     * already in the environment, so the real admin token is put back over this sentinel before the
-     * member process ever starts. That defeat applies to <em>every</em> name {@code secrets.sh}
-     * exports, and no launcher-side overlay can win against it.
-     *
-     * <p>So this constant is not the control on its own — {@link #MEMBER_MARKER} is the other half.
-     * Keeping the sentinel is still worth it: it is correct for any peer kind whose pane does not
-     * start a login shell, and it makes the intent explicit at the one place every adapter passes.
-     */
-    private static final String BLOCKED_CREDENTIAL_SENTINEL =
-            "blocked-by-bridged-cb596-see-gitea-issue-82";
-
-    /**
-     * CB-592: marks a pane as a bridged member so a shell startup file can decline to export
-     * operator-only credentials into it (gitea issue #77).
-     *
-     * <p>This name is deliberately one that {@code secrets.sh} never exports, which is exactly why
-     * it survives the login shell that wipes {@link #BLOCKED_GITEA_ACCESS_TOKEN}. The mechanism is
-     * measured, not assumed: {@code GITEA_TOKEN} is injected the same way, is absent from a login
-     * shell of its own, and was observed set inside a live member pane.
-     *
-     * <p>It is a no-op until the operator guards the export, which is a one-line change in a file
-     * this repo does not own and must not edit unasked:
-     *
-     * <pre>{@code
-     * [ -n "${BRIDGED_MEMBER:-}" ] || export GITEA_ACCESS_TOKEN=...
-     * }</pre>
-     *
-     * <p>Setting the marker now costs nothing and means that edit is the whole remaining fix.
-     */
-    static final String MEMBER_MARKER = "BRIDGED_MEMBER";
-
-    /**
-     * Seed a worker's environment (CB-511): the daemon's own {@code PATH}, then the profile's
-     * {@code env:} entries, then the CB-592 admin-token shadow.
-     *
-     * <p>Why this exists: bridged passes herdr an explicit env map, and herdr merges it into
-     * <em>its own</em> process environment. So before this, a worker inherited whatever PATH the
-     * herdr server happened to be started with — on this host, one from weeks earlier with no JDK
-     * and no Maven, which left workers unable to run the build they were being asked to run. The
-     * worker's toolchain must follow from configuration, not from how a long-lived daemon was
-     * launched.
-     *
-     * <p>Adapter-specific variables are layered on top of this by {@code buildLaunch} and therefore
-     * win. That ordering is deliberate and load-bearing: it stops a profile's {@code env:} from
-     * overriding {@code ANTHROPIC_BASE_URL} and slipping past {@link
-     * dev.ltms.bridged.guard.SubscriptionGuard}, which is checked against the profile's
-     * {@code baseUrl} and nothing else.
-     *
-     * <p>The CB-596 credential shadow and the CB-592 marker are put in <em>last</em>, after the
-     * profile's own {@code env:}, so no profile — present or future — can restore a blocked
-     * credential, or hide that the pane is a member, by naming either in config. This is the one
-     * place both are applied: every {@code buildLaunch} in every adapter calls this first, so a new
-     * profile, and a peer kind not yet written, gets them for free.
-     */
-    protected Map<String, String> baseEnv(BridgedConfig.Profile cfg) {
-        Map<String, String> workerEnv = new LinkedHashMap<>();
-        String path = env.apply("PATH");
-        if (path != null && !path.isBlank()) {
-            workerEnv.put("PATH", path);
-        }
-        if (cfg != null && cfg.env() != null) {
-            workerEnv.putAll(cfg.env());
-        }
-        applyMemberCredentialPolicy(workerEnv);
-        workerEnv.put(MEMBER_MARKER, "1");
-        return workerEnv;
-    }
-
-    /**
-     * CB-596: shadow every configured {@code memberCredentials.known} name that is not also
-     * {@code allow}-ed, replacing CB-592's single hardcoded {@code GITEA_ACCESS_TOKEN} name (gitea
-     * issue #82). An allow-listed name is deliberately left unmentioned here — see {@link
-     * #BLOCKED_CREDENTIAL_SENTINEL}'s javadoc for why an overlay entry is the only way to shadow an
-     * inherited value, which is exactly why an allowed name must get NO entry: any entry at all,
-     * blank or not, risks overriding the real value the pane needs.
-     *
-     * <p>No {@code memberCredentials} configured — the supplier is {@code null}, or it resolves to
-     * one whose {@code known} list is empty — shadows nothing. This is a real, config-driven gap
-     * (see {@link BridgedConfig.MemberCredentials}'s javadoc), not a safe default: deny-by-default
-     * only defends names the operator has actually enumerated in {@code known}.
-     */
-    private void applyMemberCredentialPolicy(Map<String, String> workerEnv) {
-        BridgedConfig.MemberCredentials creds = memberCredentials == null ? null : memberCredentials.get();
-        if (creds == null) {
-            return;
-        }
-        for (String name : creds.blockedSet()) {
-            workerEnv.put(name, BLOCKED_CREDENTIAL_SENTINEL);
-        }
-        logCredentialGap(creds);
-    }
-
-    /** Credential-shaped env var name heuristic for {@link #logCredentialGap} — case-insensitive. */
-    private static final Pattern CREDENTIAL_SHAPED_NAME =
-            Pattern.compile("(?i).*(TOKEN|SECRET|_KEY|APIKEY|PASSWORD|CREDENTIAL|AUTH).*");
-
-    /** Guards {@link #logCredentialGap} to one WARN per launcher instance, not one per spawn. */
-    private final AtomicBoolean credentialGapLogged = new AtomicBoolean();
-
-    /**
-     * CB-596 criterion 4: a credential-shaped host env var name on neither {@code known} nor
-     * {@code allow} is not silently allowed — it is reported. {@link #hostEnvNames} enumerates the
-     * daemon's own environment (see that field's javadoc for why the daemon's env is read rather
-     * than the spawned pane's, which the daemon has no channel to inspect at spawn time); this logs
-     * every such NAME, at WARN, at most once per launcher instance — never a value, a prefix of a
-     * value, or a hash of a value, so the log itself cannot leak anything.
-     */
-    private void logCredentialGap(BridgedConfig.MemberCredentials creds) {
-        Set<String> covered = new HashSet<>(creds.known());
-        covered.addAll(creds.allow());
-        List<String> gap = hostEnvNames.get().stream()
-                .filter(name -> CREDENTIAL_SHAPED_NAME.matcher(name).matches())
-                .filter(name -> !covered.contains(name))
-                .sorted()
-                .toList();
-        if (gap.isEmpty()) {
-            return;
-        }
-        if (credentialGapLogged.compareAndSet(false, true)) {
-            log.warn("memberCredentials gap: {} credential-shaped env var name(s) are on neither "
-                            + "known: nor allow: — every member pane inherits them UNBLOCKED — {}. "
-                            + "Add each to memberCredentials.known (blocked by default) or .allow "
-                            + "(if a member legitimately needs it).",
-                    gap.size(), gap);
-        }
-    }
-
-    /** Defensive copy of {@code argv} plus room to append launch flags. */
-    protected static List<String> mutableArgv(List<String> argv) {
-        return new ArrayList<>(argv);
-    }
-
-    /**
-     * Uninterruptible sleep — the production {@link #sleeper}. Tests supply their own no-op /
-     * fast-faking sleeper so they never real-sleep.
-     */
-    protected static void sleepUninterruptibly(long ms) {
-        try {
-            Thread.sleep(ms);
-        } catch (InterruptedException e) {
-            Thread.currentThread().interrupt();
-            // preserve the interrupt flag but continue — poll loops should not be aborted by an
-            // interrupt that was not meant for them.
-        }
-    }
-}

bridged/src/main/java/dev/ltms/bridged/member/OpenCodeLauncher.java

@@ -1,496 +0,0 @@
-package dev.ltms.bridged.member;
-
-import com.fasterxml.jackson.databind.ObjectMapper;
-import com.fasterxml.jackson.databind.node.ObjectNode;
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.herdr.Agent;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.peer.Capability;
-import dev.ltms.bridged.peer.CharterReceipt;
-import dev.ltms.bridged.peer.PeerHandle;
-import dev.ltms.bridged.peer.SpawnRequest;
-
-import java.io.IOException;
-import java.io.UncheckedIOException;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.EnumSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Set;
-import java.util.function.Function;
-import java.util.function.LongSupplier;
-import java.util.function.Supplier;
-
-/**
- * The {@link HerdrPeerLauncher} adapter for <strong>opencode</strong> — an open-source,
- * provider-agnostic terminal coding agent. Its whole reason for existing is to prove the
- * {@code PeerLauncher} SPI is genuinely provider-neutral: opencode shares none of Claude Code's
- * private launch seams, yet reuses every line of shared transport in the base (tab/pane placement,
- * the CB-306 readiness gate, unique naming + CB-117 reap, teardown, listing, cwd).
- *
- * <p>The divergences from {@link ClaudeCodeLauncher}, all confined to {@link #buildLaunch}:
- * <ul>
- *   <li><strong>No subscription boundary.</strong> opencode carries no {@code ANTHROPIC_BASE_URL}
- *       and there is no {@link dev.ltms.bridged.guard.SubscriptionGuard} — the guard is a
- *       Claude-private concern, not part of the SPI. opencode reads the operator's own provider
- *       credentials from its global {@code auth.json}; the bridge injects none.</li>
- *   <li><strong>File-based MCP mount + instructions.</strong> opencode has no inline
- *       {@code --mcp-config}/{@code --append-system-prompt}. Instead the bridge writes an ephemeral
- *       {@code opencode.json} that declares the bridge as a {@code remote} MCP server and lists a
- *       member-charter file under {@code instructions}, then points the worker at it with
- *       {@code OPENCODE_CONFIG}. This is the one place the launcher touches disk — Claude never did.</li>
- *   <li><strong>Model as a flag.</strong> the {@code provider/model} selector is passed as
- *       {@code -m}, not an env var.</li>
- *   <li><strong>{@code opencode} name prefix</strong> so reap matches {@code opencode-*} panes and
- *       never another adapter's.</li>
- * </ul>
- */
-public final class OpenCodeLauncher extends HerdrPeerLauncher {
-
-    /** Label prefix for this adapter's herdr agent names (drives naming + orphan reap). */
-    private static final String NAME_PREFIX = "opencode";
-
-    /** Writer for the generated {@code opencode.json}. */
-    private static final ObjectMapper JSON = new ObjectMapper();
-
-    /** Root under which per-spawn opencode config dirs are created (injectable for tests). */
-    private final Path configRoot;
-
-    /**
-     * Session discovery against opencode's on-disk storage ({@link OpenCodeSessionDiscovery}) —
-     * the one seam that knows opencode's private session-file layout. Its root is injectable for
-     * tests so they never touch the operator's real {@code ~/.local/share/opencode}.
-     */
-    private final OpenCodeSessionDiscovery discovery;
-
-    /**
-     * Production constructor — disables the spawn-ready gate ({@code spawnReadyTimeoutMs == 0}) so it
-     * matches the legacy non-blocking spawn semantics. Config dirs are created under the JVM temp dir.
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env) {
-        this(agents, spaces, profiles, defaultProfile, env, 0,
-                System::currentTimeMillis, () -> sleepUninterruptibly(300),
-                defaultConfigRoot(), defaultDiscoveryRoot());
-    }
-
-    /**
-     * Production constructor with the spawn-ready gate enabled. Polls {@code agents.status()} until
-     * the pane reports an injectable state or {@code spawnReadyTimeoutMs} elapses.
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs, long spawnReadyPollMs) {
-        this(agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, spawnReadyPollMs, null);
-    }
-
-    /**
-     * Production constructor carrying the fleet-wide tab-label template (CB-557). The template comes
-     * from {@code fleet.tabLabel}, which a profile cannot know because it names the member's
-     * <em>role</em>; a profile may still override it with its own {@code tabLabel}.
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs, long spawnReadyPollMs,
-                            Supplier<BridgedConfig.Fleet> fleet) {
-        this(agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                System::currentTimeMillis, () -> sleepUninterruptibly(spawnReadyPollMs),
-                defaultConfigRoot(), defaultDiscoveryRoot(), fleet);
-    }
-
-    /**
-     * Production constructor, plus the CB-596 {@code memberCredentials} policy supplier.
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs, long spawnReadyPollMs,
-                            Supplier<BridgedConfig.Fleet> fleet,
-                            Supplier<BridgedConfig.MemberCredentials> memberCredentials) {
-        this(agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                System::currentTimeMillis, () -> sleepUninterruptibly(spawnReadyPollMs),
-                defaultConfigRoot(), defaultDiscoveryRoot(), fleet, memberCredentials);
-    }
-
-    /**
-     * Full testability constructor. Every injectable collaborator is explicit so unit tests supply a
-     * fake clock ({@code nowMillis}), poll-loop wait ({@code sleeper}), and a temp {@code configRoot}
-     * they can inspect the generated {@code opencode.json}/charter under.
-     *
-     * @param agents              herdr agent control (start, status, close)
-     * @param spaces              workspace / tab control (ensure, create, close)
-     * @param profiles            configured worker profiles
-     * @param defaultProfile      profile a no-argument spawn uses (nullable)
-     * @param env                 host env lookup (injectable for tests)
-     * @param spawnReadyTimeoutMs max ms to wait for injectable state (0 disables the gate)
-     * @param nowMillis           monotonic clock source (e.g. {@code System::currentTimeMillis})
-     * @param sleeper             sleep/wait hook (encodes the poll interval; never called when the
-     *                            gate is disabled)
-     * @param configRoot          existing directory under which per-spawn config dirs are created
-     * @param discoveryRoot       opencode's on-disk storage root to scan for session records
-     *                            (injectable for tests; opencode's layout is matched at
-     *                            {@link OpenCodeSessionDiscovery})
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs,
-                            LongSupplier nowMillis, Runnable sleeper,
-                            Path configRoot, Path discoveryRoot) {
-        this(agents, spaces, profiles, defaultProfile, env, spawnReadyTimeoutMs,
-                nowMillis, sleeper, configRoot, discoveryRoot, null);
-    }
-
-    /**
-     * Full testability constructor, plus the fleet-wide tab-label template (CB-557).
-     *
-     * @param fleet live fleet config, read once for each spawn
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs,
-                            LongSupplier nowMillis, Runnable sleeper,
-                            Path configRoot, Path discoveryRoot,
-                            Supplier<BridgedConfig.Fleet> fleet) {
-        super(NAME_PREFIX, agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, fleet);
-        this.configRoot = configRoot;
-        this.discovery = new OpenCodeSessionDiscovery(discoveryRoot);
-    }
-
-    /**
-     * Full testability constructor, plus the CB-596 {@code memberCredentials} policy supplier.
-     */
-    public OpenCodeLauncher(AgentControl agents, WorkspaceControl spaces,
-                            Map<String, BridgedConfig.Profile> profiles, String defaultProfile,
-                            Function<String, String> env,
-                            long spawnReadyTimeoutMs,
-                            LongSupplier nowMillis, Runnable sleeper,
-                            Path configRoot, Path discoveryRoot,
-                            Supplier<BridgedConfig.Fleet> fleet,
-                            Supplier<BridgedConfig.MemberCredentials> memberCredentials) {
-        super(NAME_PREFIX, agents, spaces, profiles, defaultProfile, env,
-                spawnReadyTimeoutMs, nowMillis, sleeper, fleet, memberCredentials);
-        this.configRoot = configRoot;
-        this.discovery = new OpenCodeSessionDiscovery(discoveryRoot);
-    }
-
-    private static Path defaultConfigRoot() {
-        return Path.of(System.getProperty("java.io.tmpdir"));
-    }
-
-    /** The default opencode storage root: {@code ~/.local/share/opencode} (the XDG data dir). */
-    private static Path defaultDiscoveryRoot() {
-        return Path.of(System.getProperty("user.home"), ".local", "share", "opencode");
-    }
-
-    /**
-     * {@inheritDoc}
-     *
-     * <p>Builds the opencode launch: no {@code ANTHROPIC_*} and no guard (opencode reads its own
-     * provider credentials); when the profile mounts the bridge MCP or has a member charter,
-     * generate an ephemeral {@code opencode.json} (remote MCP server + member-charter instructions)
-     * and point the worker at it via {@code OPENCODE_CONFIG}; carry the parity-neutral git-forge
-     * grant; and select the model with {@code -m}.
-     */
-    @Override
-    protected Launch buildLaunch(BridgedConfig.Profile cfg, LaunchSpec spec) {
-        Map<String, String> workerEnv = baseEnv(cfg);
-        // A config file is needed for the bridge MCP mount, a member charter, or a pinned endpoint (CB-508).
-        if (cfg.hasMcp() || spec.charter() != null || hasCustomProvider(cfg)) {
-            workerEnv.put("OPENCODE_CONFIG", writeConfig(cfg, spec.charter()).toString());
-        }
-        applyGitToken(workerEnv, cfg);
-        return new Launch(workerEnv,
-                argvWithResume(argvWithModel(argvWithAuto(cfg), cfg), spec.resumeSessionId()));
-    }
-
-    /**
-     * True when this profile pins its own OpenAI-compatible endpoint (CB-508) rather than using
-     * whatever provider opencode resolves by default.
-     *
-     * <p>Note this reuses {@code baseUrl}, the same field the Claude adapter injects as
-     * {@code ANTHROPIC_BASE_URL} — but it does <em>not</em> go through {@code SubscriptionGuard}.
-     * That asymmetry is deliberate and safe: the guard exists to stop a worker borrowing the
-     * primary's Anthropic subscription, and an opencode process has no Anthropic credential path
-     * at all. Pointing it at a local vLLM cannot leak the subscription.
-     */
-    private static boolean hasCustomProvider(BridgedConfig.Profile cfg) {
-        return cfg.baseUrl() != null && !cfg.baseUrl().isBlank();
-    }
-
-    /**
-     * The launch argv plus the unconditional {@code --auto} flag, which auto-approves the
-     * permissions opencode does not explicitly deny. It is unconditional, not a preference: a
-     * spawned peer has no human at its pane — the bridge spawned it — so one that stops at an
-     * approval prompt is a wedged agent, indistinguishable from a legitimate mid-turn wait and
-     * unable to end its turn with {@code bridge_reply}. opencode's own help calls this
-     * "dangerous!", but the blast radius here is already bounded by design: a worker runs in its
-     * own git worktree on its own branch, is off-subscription, and cannot merge — the lead is the
-     * gate.
-     */
-    private List<String> argvWithAuto(BridgedConfig.Profile cfg) {
-        List<String> argv = mutableArgv(cfg.argv());
-        argv.add("--auto");
-        return argv;
-    }
-
-    /**
-     * The launch argv plus, on a resumed spawn, opencode's {@code -s <id>} flag to continue a prior
-     * conversation by its session id. {@code -s, --session <id>} resumes an existing session; on a
-     * fresh spawn (no resume target) no flag is added, letting opencode start a brand-new session.
-     * The id comes from the base launch spec.
-     */
-    private List<String> argvWithResume(List<String> argv, String id) {
-        if (id == null || id.isBlank()) {
-            return argv;
-        }
-        List<String> withResume = mutableArgv(argv);
-        withResume.add("-s");
-        withResume.add(id);
-        return withResume;
-    }
-
-    /** The launch argv plus, when a model is configured, the opencode {@code -m provider/model} flag. */
-    private List<String> argvWithModel(List<String> argv, BridgedConfig.Profile cfg) {
-        if (cfg.model() != null && !cfg.model().isBlank()) {
-            argv.add("-m");
-            argv.add(cfg.model());
-        }
-        return argv;
-    }
-
-    /**
-     * Write an ephemeral {@code opencode.json} (and the member-charter file it references) into a
-     * fresh per-spawn directory under {@link #configRoot}, and return the config file's path for
-     * {@code OPENCODE_CONFIG}. The dir is unique per spawn so concurrent workers never race on it;
-     * it is best-effort cleaned on JVM exit (worker config is disposable — regenerated every spawn).
-     */
-    private Path writeConfig(BridgedConfig.Profile cfg, String charterText) {
-        try {
-            Path dir = Files.createTempDirectory(configRoot, "bridged-opencode-");
-            dir.toFile().deleteOnExit();
-
-            ObjectNode root = JSON.createObjectNode();
-            root.put("$schema", "https://opencode.ai/config.json");
-            // CB-523, opencode side: a worker that runs out of context dies mid-turn, and its reply
-            // — the entire point of the turn — is lost with it. Auto-compaction is therefore not an
-            // operator preference for a bridged worker, it is a condition of the turn contract.
-            //
-            // Stated deliberately even though it is redundant today: OPENCODE_CONFIG is MERGED over
-            // ~/.config/opencode/config.json rather than replacing it, so a worker already inherits
-            // an `auto: true` set at home. We do not want that inheritance to be what the guarantee
-            // rests on — the home file is outside this repo, differs per machine, and is not ours.
-            //
-            // Know the cost before removing it: this key WINS over the home config (verified — an
-            // OPENCODE_CONFIG value overrides the home value, it does not defer to it), so an
-            // operator who sets `compaction.auto: false` at home cannot turn it off for bridged
-            // workers. That is the intended trade for peers we spawn and whose turns we must land;
-            // if per-profile control is ever wanted, add a profile knob rather than dropping this.
-            root.putObject("compaction").put("auto", true);
-
-            if (charterText != null) {
-                Path charter = dir.resolve("member-charter.md");
-                Files.writeString(charter, charterText);
-                charter.toFile().deleteOnExit();
-
-                root.putArray("instructions").add(charter.toAbsolutePath().toString());
-            }
-
-            if (cfg.hasMcp()) {
-                ObjectNode bridge = root.putObject("mcp").putObject("bridge");
-                bridge.put("type", "remote");
-                bridge.put("url", cfg.mcpUrl());
-                bridge.put("enabled", true);
-            }
-            if (hasCustomProvider(cfg)) {
-                addCustomProvider(root, cfg);
-            }
-
-            Path cfgFile = dir.resolve("opencode.json");
-            // Built with Jackson rather than string concatenation: the provider block is nested and
-            // carries operator-supplied values (URL, model id, api key), so escaping must be real.
-            Files.writeString(cfgFile, JSON.writerWithDefaultPrettyPrinter().writeValueAsString(root));
-            cfgFile.toFile().deleteOnExit();
-            return cfgFile;
-        } catch (IOException e) {
-            throw new UncheckedIOException(
-                    "cannot write opencode config for profile " + cfg.profile(), e);
-        }
-    }
-
-    /**
-     * Declare a custom OpenAI-compatible provider so the worker talks to a pinned endpoint (a local
-     * vLLM, say) instead of opencode's default gateway (CB-508).
-     *
-     * <p>The provider id comes from the {@code provider/model} selector in {@code model:}, so one
-     * field drives both the declaration and the {@code -m} flag and they cannot drift apart.
-     */
-    private void addCustomProvider(ObjectNode root, BridgedConfig.Profile cfg) {
-        String[] parts = splitModelSelector(cfg);
-        String providerId = parts[0];
-        String modelId = parts[1];
-
-        ObjectNode provider = root.putObject("provider").putObject(providerId);
-        provider.put("npm", "@ai-sdk/openai-compatible");
-        provider.put("name", providerId + " (bridged)");
-
-        ObjectNode options = provider.putObject("options");
-        options.put("baseURL", openAiBaseUrl(cfg.baseUrl()));
-        // vLLM and friends usually ignore the key, but the AI SDK still requires a non-empty one.
-        String token = resolveEnv(cfg.tokenEnv());
-        options.put("apiKey", (token == null || token.isBlank()) ? "bridged-local-noauth" : token);
-
-        provider.putObject("models").putObject(modelId).put("name", modelId);
-    }
-
-    /**
-     * Split {@code model:} into its {@code provider} and {@code model} halves. A pinned endpoint
-     * needs both, so a bare model name is rejected loudly rather than silently falling back to the
-     * default gateway — a worker quietly talking to the wrong endpoint is the failure this avoids.
-     */
-    private static String[] splitModelSelector(BridgedConfig.Profile cfg) {
-        String model = cfg.model();
-        int slash = model == null ? -1 : model.indexOf('/');
-        if (model == null || model.isBlank() || slash <= 0 || slash == model.length() - 1) {
-            throw new IllegalArgumentException(
-                    "profile " + cfg.profile() + " sets baseUrl (a pinned opencode endpoint) so"
-                            + " model: must be \"<provider>/<model>\", e.g."
-                            + " \"local-vllm/deepseek-v4-flash\"; got "
-                            + (model == null ? "null" : '"' + model + '"'));
-        }
-        return new String[]{model.substring(0, slash), model.substring(slash + 1)};
-    }
-
-    /**
-     * The OpenAI-compatible base URL for {@code baseUrl}. A bare {@code host:port} gets {@code /v1}
-     * appended (where these servers put the API); a URL that already carries a path is taken as-is,
-     * so an endpoint mounted somewhere unusual is still reachable.
-     */
-    private static String openAiBaseUrl(String baseUrl) {
-        String trimmed = baseUrl.trim();
-        while (trimmed.endsWith("/")) {
-            trimmed = trimmed.substring(0, trimmed.length() - 1);
-        }
-        int schemeEnd = trimmed.indexOf("://");
-        String afterScheme = schemeEnd < 0 ? trimmed : trimmed.substring(schemeEnd + 3);
-        return afterScheme.contains("/") ? trimmed : trimmed + "/v1";
-    }
-
-    /** Add lazy on-disk session discovery to the base handle. */
-    @Override
-    public PeerHandle spawn(SpawnRequest req) {
-        PeerHandle inner = super.spawn(req);
-        return new SessionAwareHandle(inner, discovery, effectiveCwd(req));
-    }
-
-    /**
-     * A {@link PeerHandle} that delegates everything to the base's worker handle but resolves
-     * {@link #agentSessionId()} lazily through opencode session discovery. Delegate-only, so the
-     * base's id/terminalId/profile semantics (CB-519's host-unique routing key, herdr coordinates)
-     * are untouched — only the opencode-specific identity answer is added. {@code sessionName()}
-     * stays null: opencode has no display-name seam, so the logical name lives only in the bridge's
-     * roster (see the SESSION_NAME capability).
-     */
-    private static final class SessionAwareHandle implements PeerHandle {
-        private final PeerHandle delegate;
-        private final OpenCodeSessionDiscovery discovery;
-        private final String cwd;
-
-        SessionAwareHandle(PeerHandle delegate, OpenCodeSessionDiscovery discovery, String cwd) {
-            this.delegate = delegate;
-            this.discovery = discovery;
-            this.cwd = cwd;
-        }
-
-        @Override
-        public String id() {
-            return delegate.id();
-        }
-
-        @Override
-        public String terminalId() {
-            return delegate.terminalId();
-        }
-
-        @Override
-        public String profile() {
-            return delegate.profile();
-        }
-
-        @Override
-        public String sessionName() {
-            return delegate.sessionName();
-        }
-
-        @Override
-        public String agentSessionId() {
-            // Lazy + retried, never a spawn-time blocker: opencode writes the session record only
-            // when the session is first persisted, so null here is the correct interim answer and
-            // the caller re-calls later (each call re-scans, picking up a record that has since
-            // appeared).
-            return discovery.sessionIdForDirectory(cwd);
-        }
-
-        @Override
-        public CharterReceipt charterReceipt() {
-            return delegate.charterReceipt();
-        }
-    }
-
-    // --- Agent-returning convenience spawns (used by callers/tests that want the herdr Agent) ---
-
-    /** Spawn a worker for the default profile in the resolved default cwd. */
-    public Agent spawn() {
-        return spawnInternal(null, null, null);
-    }
-
-    /** Spawn a worker for a named profile (null → default) in the resolved default cwd. */
-    public Agent spawn(String profileName) {
-        return spawnInternal(profileName, null, null);
-    }
-
-    /** Spawn a worker for a named profile with an explicit requested/caller cwd (CB-112). */
-    public Agent spawn(String profileName, String requestedCwd, String callerCwd) {
-        return spawnInternal(profileName, requestedCwd, callerCwd);
-    }
-
-    // --- capabilities --------------------------------------------------------------------------
-
-    @Override
-    public Set<Capability> capabilities() {
-        Set<Capability> caps = EnumSet.of(Capability.MID_TURN_ASK, Capability.WORKTREE,
-                Capability.ORPHAN_REAP, Capability.SESSION_RESUME);
-        if (hasGitTokenProfile()) {
-            caps.add(Capability.SELF_PR);
-        }
-        // Deliberately NOT SESSION_NAME: opencode has no display-name flag, so the bridge's logical
-        // name can't surface in the peer's own UI — declaring the capability would hide that
-        // asymmetry rather than make it honest. For opencode the name lives only in the bridge's
-        // roster (see PeerHandle.sessionName() returning null).
-        return Set.copyOf(caps);
-    }
-
-    /** Whether any configured profile opts into a git-forge token (required for {@link Capability#SELF_PR}). */
-    private boolean hasGitTokenProfile() {
-        return profileConfigs().stream().anyMatch(BridgedConfig.Profile::hasGitToken);
-    }
-
-    // --- CB-117 reap predicate (opencode prefix), kept for direct unit testing -----------------
-
-    /**
-     * Whether {@code name} is an opencode bridge worker started by a <em>different</em> process than
-     * {@code currentNonce}. A thin {@code opencode}-prefix binding of
-     * {@link HerdrPeerLauncher#isForeignWorker(String, String, String)}.
-     */
-    static boolean isForeignWorker(String name, String currentNonce) {
-        return HerdrPeerLauncher.isForeignWorker(NAME_PREFIX, name, currentNonce);
-    }
-}

bridged/src/main/java/dev/ltms/bridged/member/OpenCodeSessionDiscovery.java

@@ -1,122 +0,0 @@
-package dev.ltms.bridged.member;
-
-import com.fasterxml.jackson.databind.JsonNode;
-import com.fasterxml.jackson.databind.ObjectMapper;
-
-import java.io.IOException;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.stream.Stream;
-
-/**
- * Resolves the opencode session id for a bridged worker from opencode's on-disk storage — the
- * only place this adapter touches opencode's private layout, and deliberately the <em>only</em>
- * class that does.
- *
- * <p><strong>Why this is isolated behind one seam.</strong> The layout is version-coupled and not a
- * stable contract: opencode writes one JSON file per session under
- * {@code <storageRoot>/session/<projectID>/<ses_*.json>}, and each record carries a
- * {@code "version"} field (e.g. {@code "1.1.31"}), so the exact directory shape, file naming, and
- * field names can move between opencode releases. opencode also ships a headless HTTP server that
- * may supersede file scanning entirely. Everything this adapter knows about that private storage —
- * its shape, naming, and field names — lives here, so a layout change, or a switch to the HTTP
- * server, changes exactly one class and nothing in {@link OpenCodeLauncher}.
- *
- * <p>The determinism that makes this useful is structural, not a guess: every bridged worker runs
- * in its own unique git worktree, so the record's {@code directory} (its project root) equals the
- * worker's cwd identifies <em>its</em> session unambiguously. We match on {@code directory} rather
- * than diffing {@code opencode session list} before/after — that races under concurrent spawns, and
- * the CLI listing does not even show the directory.
- *
- * <p>All reads are best-effort and never throw: a missing or unreadable storage root, a record that
- * fails to parse, or a directory with no record yet all yield {@code null}, and the caller (the
- * session handle) treats that as "identity not resolved yet" and retries later.
- */
-final class OpenCodeSessionDiscovery {
-
-    private final Path storageRoot;   // e.g. ~/.local/share/opencode (injectable for tests)
-    private final ObjectMapper json;
-
-    OpenCodeSessionDiscovery(Path storageRoot) {
-        this.storageRoot = storageRoot;
-        this.json = new ObjectMapper();
-    }
-
-    /**
-     * The opencode session id whose record references {@code directory} (the worker's cwd), or
-     * {@code null} when no record matches yet. When several records share the directory — e.g.
-     * repeated spawns into the same worktree — the <em>most recently modified</em> one wins: it is
-     * the session the pane most likely corresponds to.
-     *
-     * <p>Never throws: a missing {@code storageRoot}, an unreadable/malformed record, or a
-     * directory that has not been persisted yet all resolve to {@code null} rather than failing a
-     * spawn. A bridged worker's session record is written lazily (when the session is first
-     * persisted), so {@code null} here is the normal answer right after the pane is ready, and the
-     * caller retries later.
-     *
-     * @param directory the worker's cwd, as resolved for this spawn
-     * @return the matching session id, or {@code null} if none is known yet
-     */
-    String sessionIdForDirectory(String directory) {
-        if (directory == null || directory.isBlank()) {
-            return null;
-        }
-        Path sessionRoot = storageRoot.resolve("session");
-        if (!Files.isDirectory(sessionRoot)) {
-            return null;
-        }
-        String best = null;
-        long bestMtime = Long.MIN_VALUE;
-        try (Stream<Path> projectDirs = Files.list(sessionRoot)) {
-            for (Path projectDir : projectDirs.filter(Files::isDirectory).toList()) {
-                try (Stream<Path> records = Files.list(projectDir)) {
-                    for (Path record : records.toList()) {
-                        String id = matchId(record, directory);
-                        if (id == null) {
-                            continue;
-                        }
-                        long mtime = lastModifiedEpochMillis(record);
-                        if (mtime > bestMtime) {
-                            bestMtime = mtime;
-                            best = id;
-                        }
-                    }
-                } catch (IOException ignored) {
-                    // one project dir unreadable — skip it; another may still match
-                }
-            }
-        } catch (IOException ignored) {
-            // storage root vanished or became unreadable — "no session known yet"
-            return null;
-        }
-        return best;
-    }
-
-    /**
-     * The record's session id when it references {@code directory}, else {@code null}. A record
-     * that is not JSON, lacks {@code id}/{@code directory}, or points at a different directory is
-     * simply not our session; a malformed one is skipped, never fatal.
-     */
-    private String matchId(Path record, String directory) {
-        try {
-            JsonNode node = json.readTree(record.toFile());
-            JsonNode id = node == null ? null : node.get("id");
-            JsonNode dir = node == null ? null : node.get("directory");
-            if (id == null || dir == null || !directory.equals(dir.asText())) {
-                return null;
-            }
-            return id.asText();
-        } catch (IOException e) {
-            return null;
-        }
-    }
-
-    /** The record's last-modified epoch ms, or {@code Long.MIN_VALUE} if unreadable (never wins). */
-    private static long lastModifiedEpochMillis(Path record) {
-        try {
-            return Files.getLastModifiedTime(record).toMillis();
-        } catch (IOException e) {
-            return Long.MIN_VALUE;
-        }
-    }
-}

bridged/src/main/java/dev/ltms/bridged/peer/PeerLauncher.java

@@ -1,107 +0,0 @@
-package dev.ltms.bridged.peer;
-
-import java.util.List;
-import java.util.Set;
-
-/**
- * SPI for materializing a connected peer — the only way the bridge core creates or tears down
- * a peer process. Every launcher is a first-party, in-tree adapter selected by (future) profile
- * config; today's single adapter is the {@code ClaudeCodeLauncher} / Claude Code over herdr.
- *
- * <p>The core delegates spawn and teardown to this interface without knowing how the peer is set
- * up. Environment variables, CLI flags, subscription guards, transport (herdr tab/pane) layout,
- * and naming conventions are all adapter-private — the core sees only the returned
- * {@link PeerHandle} whose {@code id()} is the registry/routing key.
- *
- * <p>The interface is a superset of what {@code SessionManager} and {@code Bridged.main} call
- * on the concrete launcher today.
- */
-public interface PeerLauncher {
-
-    /**
-     * The set of {@link Capability capabilities} this launcher declares. A peer whose profile
-     * opts into a git-forge token should include {@link Capability#SELF_PR}; the base set for
-     * the Claude Code herdr adapter is always {@code MID_TURN_ASK, WORKTREE, ORPHAN_REAP}.
-     */
-    Set<Capability> capabilities();
-
-    /**
-     * The capabilities of the adapter that {@code profileName} resolves to (null/blank → the
-     * default profile, the same resolution {@link #spawn} uses). Distinct from {@link
-     * #capabilities()}, which unions every configured adapter: a caller that must know whether
-     * <em>this</em> profile's backend supports a capability — e.g. {@link Capability#SESSION_RESUME}
-     * before honoring {@link SpawnRequest#resumeSessionId()} — needs the per-profile answer, not
-     * the fleet-wide union, or a mixed fleet could OK a resume that lands on a non-supporting
-     * adapter (CB-584).
-     *
-     * @throws IllegalArgumentException if the profile is unknown and no default is configured
-     */
-    Set<Capability> capabilitiesFor(String profileName);
-
-    /**
-     * {@code profileName}/requestedCwd null/blank → default resolution. Returns after the peer
-     * process is live (env + argv + placement complete). Never returns {@code null}.
-     *
-     * @param req the spawn parameters (profile, requested cwd, caller cwd)
-     * @return a handle whose {@link PeerHandle#id()} is the registry/routing key
-     * @throws IllegalArgumentException if the profile is unknown and no default is configured
-     */
-    PeerHandle spawn(SpawnRequest req);
-
-    /**
-     * The configured worker profile names — the set of names {@code spawn(profileName)} accepts.
-     */
-    Set<String> profiles();
-
-    /**
-     * The profile a no-argument {@link #spawn(SpawnRequest)} uses, or {@code null} if none is configured.
-     */
-    String defaultProfile();
-
-    /**
-     * Resolve the effective working directory for a spawn {@code req} without actually spawning.
-     * Resolution order: requestedCwd → profile cwd → callerCwd → daemon cwd.
-     *
-     * @return the resolved absolute path, never null/blank
-     */
-    String effectiveCwd(SpawnRequest req);
-
-    /**
-     * The parity-overlay file list for {@code profileName} (default list when unset). Used by
-     * worktree provisioning to copy config files into the isolated checkout before spawning.
-     */
-    List<String> parityOverlay(String profileName);
-
-    /**
-     * The set of all agents this launcher currently tracks, transport-specific. Each element
-     * exposes at minimum a pane-like {@code id()} matching this launcher's {@link PeerHandle}
-     * scheme, plus transport-level status. Callers merge this set with the session registry to
-     * build a live roster view.
-     */
-    List<?> list();
-
-    /**
-     * Reap orphaned peers left behind by a prior daemon process. Only peers whose naming scheme
-     * matches this launcher's and whose nonce differs from the current process are eligible.
-     * Best-effort: a failure to list or to stop any one peer is logged and never aborts startup.
-     *
-     * @return the number of orphaned peers reaped
-     */
-    int reapOrphanWorkers();
-
-    /**
-     * Tear a peer down by its registry/routing key ({@link PeerHandle#id()}). Tolerates an
-     * already-gone peer. Also cleans up launcher-private resources (e.g. empty dedicated tabs)
-     * when safe to do so.
-     */
-    void stop(String id);
-
-    /**
-     * Discard the context of the peer identified by {@code id}. Implementations must bypass normal
-     * bridge delivery/turn accounting. Unsupported peer kinds return {@code false} without sending
-     * a guessed command.
-     *
-     * @return {@code true} when a reset was sent and its status transition must settle before reuse
-     */
-    boolean clearContext(String id);
-}

bridged/src/main/java/dev/ltms/bridged/session/GitWorktrees.java

@@ -1,510 +0,0 @@
-package dev.ltms.bridged.session;
-
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-
-import java.io.BufferedReader;
-import java.io.IOException;
-import java.io.InputStreamReader;
-import java.io.UncheckedIOException;
-import java.nio.charset.StandardCharsets;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.nio.file.StandardCopyOption;
-import java.security.SecureRandom;
-import java.util.ArrayList;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Map;
-import java.util.Optional;
-import java.util.Set;
-import java.util.concurrent.TimeUnit;
-import java.util.concurrent.atomic.AtomicLong;
-import java.util.stream.Collectors;
-
-/**
- * Production {@link Worktrees} implementation that shells {@code git} via {@link ProcessBuilder}.
- * Non-zero exits become {@link WorktreeException}. Worktree directories live under a configurable
- * root (default: a sibling {@code .bridged-worktrees} of the repo root) so they are never nested
- * inside the primary working tree.
- */
-public final class GitWorktrees implements Worktrees {
-
-    private static final Logger log = LoggerFactory.getLogger(GitWorktrees.class);
-
-    /** Project-level MCP config. Present in the repo, so every worktree would otherwise inherit the
-     *  primary's IDE server mounts (a CB-523 worker edited the primary checkout; see the isolation
-     *  javadoc). Neutralized unconditionally. */
-    private static final String MCP_CONFIG = ".mcp.json";
-
-    /** What {@link #isolateToolSurface} writes for {@code .mcp.json}: a valid, explicitly empty server map. */
-    private static final String NEUTRAL_MCP_CONFIG = "{\n  \"mcpServers\": {}\n}\n";
-
-    /** OpenCode's repo-level config. Tracked here, so it lands in every worktree, and it mounts the
-     *  primary's gitea and context7 servers with the primary's credentials. Neutralized so the worker
-     *  gets only the config its launcher writes via {@code OPENCODE_CONFIG}.
-     *
-     *  <p>The reason has changed shape and is now stronger. It used to be a crash: the file carried
-     *  {@code {file:.secrets/...}} references to gitignored files that never reached a worktree, and
-     *  opencode refuses to start on a dangling reference (CB-543). Those credentials now live in one
-     *  shell-level store and the file reads them as {@code {env:...}}, so in a worktree the reference
-     *  resolves instead of failing. That is worse, not better: a member would silently inherit the
-     *  primary's admin-scoped {@code GITEA_ACCESS_TOKEN}. A loud crash became a quiet privilege leak,
-     *  so this entry protects a boundary now rather than papering over a startup error. */
-    private static final String OPENCODE_CONFIG = "opencode.json";
-
-    /** What {@link #isolateToolSurface} writes for {@code opencode.json}: a valid, empty JSON object. */
-    private static final String NEUTRAL_OPENCODE_CONFIG = "{}\n";
-
-    /** Autoenv's repo-level config. Not tracked today, but re-landing it must stay safe: autoenv
-     *  authorizes by path, so a fresh worktree path is always unauthorized and its interactive prompt
-     *  would block every spawn — neutralize it so it can never be committed. */
-    private static final String AUTOENV_CONFIG = ".autoenv";
-
-    /** What {@link #isolateToolSurface} writes for {@code .autoenv}: a valid, empty env file. */
-    private static final String NEUTRAL_AUTOENV_CONFIG = "";
-
-    /**
-     * A tracked project config that is hostile in a provisioned worktree, and what to replace it
-     * with. {@link #file} is the repo-relative path; {@link #stub} is a neutral but VALID payload for
-     * that file's format — a malformed stub would only trade one crash for another;
-     * {@link #createIfAbsent} keeps {@code .mcp.json}'s long-standing behaviour of writing its stub
-     * even when the repo carries no such file, whereas the others are only touched when present.
-     */
-    private record WorktreeHostileConfig(String file, String stub, boolean createIfAbsent) {}
-
-    /** The worktree-hostile configs neutralized in every provisioned worktree, in order. */
-    private static final List<WorktreeHostileConfig> WORKTREE_HOSTILE_CONFIGS = List.of(
-            new WorktreeHostileConfig(MCP_CONFIG, NEUTRAL_MCP_CONFIG, true),
-            new WorktreeHostileConfig(OPENCODE_CONFIG, NEUTRAL_OPENCODE_CONFIG, false),
-            new WorktreeHostileConfig(AUTOENV_CONFIG, NEUTRAL_AUTOENV_CONFIG, false)
-    );
-
-    private final String configuredRoot;
-    private final SecureRandom random = new SecureRandom();
-    private final AtomicLong seq = new AtomicLong();
-
-    /** Default constructor: worktree root is derived per-repo as {@code <repoRoot>/../.bridged-worktrees}. */
-    public GitWorktrees() {
-        this(null);
-    }
-
-    /** @param configuredRoot nullable absolute or relative path; null/blank derives a sibling of the repo root. */
-    public GitWorktrees(String configuredRoot) {
-        this.configuredRoot = configuredRoot;
-    }
-
-    @Override
-    public String add(String repoRoot, String branch, String baseRef) {
-        String base = (baseRef == null || baseRef.isBlank()) ? "HEAD" : baseRef;
-        String nonce = nonce();
-        Path root = resolveRoot(repoRoot);
-        Path path = root.resolve(nonce);
-        try {
-            Files.createDirectories(root);
-        } catch (IOException e) {
-            throw new WorktreeException("cannot create worktree root " + root + ": " + e.getMessage(), e);
-        }
-        String wt = path.toAbsolutePath().toString();
-        log.info("adding worktree branch={} path={} base={}", branch, wt, base);
-        exec("git", "-C", repoRoot, "worktree", "add", wt, "-b", branch, base);
-        isolateToolSurface(wt);
-        return wt;
-    }
-
-    /**
-     * Neutralize the worktree's worktree-hostile project configs so a worker inherits only the tools
-     * and environment its launcher mounts (the bridge via {@code --mcp-config}, the opencode config
-     * via {@code OPENCODE_CONFIG}) — never the primary's.
-     *
-     * <p>This is unconditional, and it is not the same job as the parity overlay. The repo's own
-     * committed {@code .mcp.json} declares the primary's IDE servers, so a fresh checkout mounts them
-     * whether or not the overlay copies anything; a worker that inherits them navigates and edits
-     * through tools bound to the <em>primary's</em> IntelliJ project, which silently hands it absolute
-     * paths outside its own worktree. That is not hypothetical: a CB-523 worker made all 59 of its
-     * edits in the primary checkout while compiling its worktree, so every build it ran was of code
-     * that did not contain its changes. {@code opencode.json} is the same trap one tool over — tracked,
-     * so it lands in every worktree, and it mounts gitea and context7 with the primary's own
-     * credentials, which a member must never hold. {@code .autoenv} extends the principle to a
-     * config that is not tracked today: autoenv authorizes by path, so a fresh worktree path is always
-     * unauthorized and its interactive prompt would block every spawn, so re-landing one must be safe.
-     *
-     * <p>Where a config exists it is replaced by a valid neutral stub (an explicitly empty
-     * map/object, or an empty env file — never a deletion, which would still let a later
-     * {@code git checkout} restore the hostile copy). The {@code --skip-worktree} bit keeps the
-     * neutralized copy from ever showing up as a local modification the worker might commit. A config
-     * the repo does not carry is skipped silently — no stub is invented for a file the repo does not
-     * have, and one missing file must never fail provisioning.
-     */
-    private void isolateToolSurface(String worktreePath) {
-        Path root = Path.of(worktreePath).toAbsolutePath().normalize();
-        for (WorktreeHostileConfig cfg : WORKTREE_HOSTILE_CONFIGS) {
-            neutralize(root, worktreePath, cfg);
-        }
-    }
-
-    private void neutralize(Path root, String worktreePath, WorktreeHostileConfig cfg) {
-        Path target = root.resolve(cfg.file());
-        if (!Files.exists(target) && !cfg.createIfAbsent()) {
-            log.debug("{} absent in the worktree — skipping (repo does not carry it)", cfg.file());
-            return;
-        }
-        try {
-            Files.writeString(target, cfg.stub());
-        } catch (IOException e) {
-            throw new WorktreeException("cannot neutralize " + cfg.file() + " in the worktree: "
-                    + e.getMessage(), e);
-        }
-        if (isTracked(root, cfg.file())) {
-            exec("git", "-C", worktreePath, "update-index", "--skip-worktree", cfg.file());
-        }
-        log.debug("neutralized {} — worker tool surface is launcher-mounted only", cfg.file());
-    }
-
-    @Override
-    public void remove(String repoRoot, String worktreePath) {
-        Path p = Path.of(worktreePath);
-        if (!Files.exists(p)) {
-            log.debug("worktree {} already gone — nothing to remove", worktreePath);
-            return;
-        }
-        log.info("removing worktree {}", worktreePath);
-        exec("git", "-C", repoRoot, "worktree", "remove", "--force", worktreePath);
-    }
-
-    @Override
-    public boolean hasUncommitted(String worktreePath) {
-        // A worktree that is already gone holds no work to lose, and it must not break teardown:
-        // git -C <missing-dir> status exits non-zero and would throw where release() is mid-way
-        // through stopping a pane. Mirror remove()'s already-gone tolerance by treating it as clean.
-        Path p = Path.of(worktreePath);
-        if (!Files.exists(p)) {
-            log.debug("worktree {} already gone — nothing can be uncommitted", worktreePath);
-            return false;
-        }
-        // No --untracked-files=no: the exact shape of the work lost in CB-576 was a new file
-        // that was never added, so an untracked-only worktree is still dirty.
-        String out = exec("git", "-C", worktreePath, "status", "--porcelain");
-        return !out.isBlank();
-    }
-
-    @Override
-    public void overlayParity(String repoRoot, String worktreePath, List<String> overlay) {
-        if (overlay == null || overlay.isEmpty()) {
-            return;
-        }
-        Path srcRoot = Path.of(repoRoot).toAbsolutePath().normalize();
-        Path dstRoot = Path.of(worktreePath).toAbsolutePath().normalize();
-        for (String rel : overlay) {
-            Path src = srcRoot.resolve(rel).normalize();
-            if (!Files.exists(src)) {
-                log.debug("parity overlay source missing — skipping {}", rel);
-                continue;
-            }
-            Path dst = dstRoot.resolve(rel).normalize();
-            try {
-                Files.createDirectories(dst.getParent());
-                Files.copy(src, dst, StandardCopyOption.REPLACE_EXISTING, StandardCopyOption.COPY_ATTRIBUTES);
-                log.debug("copied parity overlay {}", rel);
-            } catch (IOException e) {
-                throw new WorktreeException("cannot copy overlay " + rel + ": " + e.getMessage(), e);
-            }
-            if (isTracked(dstRoot, rel)) {
-                exec("git", "-C", worktreePath, "update-index", "--skip-worktree", rel);
-                log.debug("marked overlay --skip-worktree {}", rel);
-            }
-        }
-    }
-
-    @Override
-    public String repoRoot(String cwd) {
-        String out = exec("git", "-C", cwd, "rev-parse", "--show-toplevel");
-        return Path.of(out.trim()).toAbsolutePath().normalize().toString();
-    }
-
-    /**
-     * CB-578 stage C, fixed by CB-587. Stages into a <em>temporary</em> index (never the worktree's
-     * real one, which the worker may still be writing to) — but that temp index is first <em>seeded</em>
-     * from the worktree's real one, rather than starting empty:
-     *
-     * <pre>
-     * cp $(git -C worktree rev-parse --git-path index) &lt;temp&gt;
-     * GIT_INDEX_FILE=&lt;temp&gt; git -C worktree add -A
-     * tree=$(GIT_INDEX_FILE=&lt;temp&gt; git -C worktree write-tree)
-     * commit=$(git -C worktree commit-tree $tree -p HEAD -m message)
-     * git -C worktree update-ref refs/wip/branch $commit
-     * </pre>
-     *
-     * A fresh empty index carries none of the real index's {@code --skip-worktree} /
-     * {@code --assume-unchanged} bits, so {@code add -A} into it stages a skip-worktree file's local
-     * on-disk content even though {@code git status --porcelain} correctly hides that file (CB-587).
-     * Seeding from the real index preserves those bits, so {@code add -A} then skips exactly what
-     * {@code git status} skips. {@code add -A} (never {@code -f}) also still respects
-     * {@code .gitignore} exactly as it would in the real index — a gitignored file staying ignored is
-     * what keeps secrets and local config out of the snapshot's tree. The temporary index file is
-     * removed afterwards regardless of outcome; the worker's real index is never opened for writing.
-     */
-    @Override
-    public Optional<String> snapshot(String worktreePath, String branch, String message) {
-        if (!Files.exists(Path.of(worktreePath))) {
-            log.debug("worktree {} already gone — nothing to snapshot", worktreePath);
-            return Optional.empty();
-        }
-        Path tempIndex;
-        try {
-            tempIndex = Files.createTempFile("bridged-wip-index-", ".tmp");
-        } catch (IOException e) {
-            throw new WorktreeException("cannot create a temporary index for snapshot: " + e.getMessage(), e);
-        }
-        Map<String, String> indexEnv = Map.of("GIT_INDEX_FILE", tempIndex.toAbsolutePath().toString());
-        try {
-            Path realIndex = resolveRealIndex(worktreePath);
-            try {
-                Files.copy(realIndex, tempIndex, StandardCopyOption.REPLACE_EXISTING);
-            } catch (IOException e) {
-                throw new WorktreeException("cannot copy the worktree's real index (" + realIndex
-                        + ") into the temporary snapshot index: " + e.getMessage(), e);
-            }
-            exec(indexEnv, "git", "-C", worktreePath, "add", "-A");
-            String tree = exec(indexEnv, "git", "-C", worktreePath, "write-tree").trim();
-            String commit = exec("git", "-C", worktreePath, "commit-tree", tree, "-p", "HEAD", "-m", message).trim();
-            exec("git", "-C", worktreePath, "update-ref", "refs/wip/" + branch, commit);
-            log.info("snapshotted worktree {} to refs/wip/{} commit={}", worktreePath, branch, commit);
-            return Optional.of(commit);
-        } finally {
-            try {
-                Files.deleteIfExists(tempIndex);
-            } catch (IOException e) {
-                log.debug("could not delete temporary snapshot index {}: {}", tempIndex, e.getMessage());
-            }
-        }
-    }
-
-    /**
-     * Resolve the path of {@code worktreePath}'s real index. Never assume {@code <worktree>/.git/index}:
-     * in a linked worktree {@code .git} is a <em>file</em> pointing at the main repo's
-     * {@code worktrees/<name>/} directory, and that is where the real per-worktree index lives.
-     * {@code git rev-parse --git-path index} resolves this correctly for both a linked worktree and
-     * the main checkout. Throws {@link WorktreeException} — same as every other failure in this
-     * class — if the command fails or the resolved path does not exist, rather than silently
-     * snapshotting from an empty index.
-     */
-    private Path resolveRealIndex(String worktreePath) {
-        String out = exec("git", "-C", worktreePath, "rev-parse", "--git-path", "index").trim();
-        Path index = Path.of(out);
-        if (!index.isAbsolute()) {
-            index = Path.of(worktreePath).resolve(index).normalize();
-        }
-        if (!Files.exists(index)) {
-            throw new WorktreeException("worktree's real index not found at resolved path " + index
-                    + " (git rev-parse --git-path index reported '" + out + "')");
-        }
-        return index;
-    }
-
-    /**
-     * One {@code refs/wip/<branch>} snapshot ref as read by {@link #listWipRefs}: its full ref name,
-     * the snapshot commit's sha, and that commit's committer time in unix millis (the age of the
-     * snapshot — a snapshot is written once and never rewritten, so the commit date is the ref's).
-     */
-    private record WipRef(String refName, String sha, long committerMillis) {
-        String branch() {
-            return refName.substring("refs/wip/".length());
-        }
-    }
-
-    @Override
-    public WipRefStats wipRefs(String repoRoot) {
-        List<WipRef> refs = listWipRefs(repoRoot);
-        long costBytes = 0;
-        for (WipRef ref : refs) {
-            costBytes += treeSize(repoRoot, ref.sha());
-        }
-        return new WipRefStats(refs.size(), costBytes);
-    }
-
-    @Override
-    public int pruneWipRefs(String repoRoot, long minAgeMillis) {
-        // The rule is documented on Worktrees#pruneWipRefs: delete only a snapshot whose tree
-        // content is already reachable from main AND that is older than minAgeMillis. Reachability
-        // is the floor that keeps a worker's last copy; the age floor keeps a just-written snapshot
-        // from being swept while a lead may still be looking at it.
-        List<WipRef> refs = listWipRefs(repoRoot);
-        if (refs.isEmpty()) {
-            return 0;
-        }
-        long nowMillis = System.currentTimeMillis();
-        // Resolve what main carries once per sweep, not once per ref.
-        Set<String> mainObjects = reachableObjectsFromMain(repoRoot);
-        int deleted = 0;
-        for (WipRef ref : refs) {
-            long ageMillis = nowMillis - ref.committerMillis();
-            if (ageMillis <= minAgeMillis) {
-                continue; // too recent — never swept, even if it looks recoverable (CB-586)
-            }
-            String tree = exec("git", "-C", repoRoot, "rev-parse", ref.sha() + "^{tree}").trim();
-            if (!mainObjects.contains(tree)) {
-                // Last copy of the snapshot's content — the worker's work exists nowhere else.
-                // Never delete automatically (CB-586 criterion 2).
-                continue;
-            }
-            exec("git", "-C", repoRoot, "update-ref", "-d", ref.refName());
-            deleted++;
-            log.info("pruned snapshot ref refs/wip/{} commit={} (age {}h): its tree is already "
-                     + "reachable from main, so the work is preserved; recover from reflog via "
-                     + "git update-ref refs/wip/{} {}",
-                    ref.branch(), ref.sha(), TimeUnit.MILLISECONDS.toHours(ageMillis),
-                    ref.branch(), ref.sha());
-        }
-        return deleted;
-    }
-
-    /**
-     * Every {@code refs/wip/*} ref (see {@link WipRef}). The committer date is read as a unix
-     * count of seconds and converted to millis. {@code %00} (NUL) separates the fields because a
-     * branch name may contain spaces.
-     */
-    private List<WipRef> listWipRefs(String repoRoot) {
-        String out = exec("git", "-C", repoRoot, "for-each-ref",
-                "--format=%(refname)%00%(objectname)%00%(committerdate:unix)", "refs/wip/");
-        List<WipRef> refs = new ArrayList<>();
-        for (String line : out.split("\\R")) {
-            if (line.isBlank()) {
-                continue;
-            }
-            String[] parts = line.split("\u0000", -1);
-            if (parts.length == 3 && !parts[1].isBlank()) {
-                refs.add(new WipRef(parts[0], parts[1], Long.parseLong(parts[2]) * 1000L));
-            }
-        }
-        return refs;
-    }
-
-    /**
-     * The set of object shas reachable from {@code main}, or an empty set when {@code main} cannot
-     * be resolved. An empty set is the safe direction: the retention sweep then concludes nothing
-     * is recoverable, so it deletes nothing — a repo with no {@code main} must never cause a
-     * worker's last copy of a snapshot to be dropped on a reachability misreading.
-     */
-    private Set<String> reachableObjectsFromMain(String repoRoot) {
-        if (exitCode("git", "-C", repoRoot, "rev-parse", "--verify", "main") != 0) {
-            log.debug("refs/wip retention: no 'main' ref in {} — treating nothing as reachable", repoRoot);
-            return Set.of();
-        }
-        String out = exec("git", "-C", repoRoot, "rev-list", "--objects", "main");
-        Set<String> objects = new HashSet<>();
-        for (String line : out.split("\\R")) {
-            if (line.isBlank()) {
-                continue;
-            }
-            int sp = line.indexOf(' ');
-            objects.add(sp < 0 ? line : line.substring(0, sp));
-        }
-        return objects;
-    }
-
-    /** Approximate cost of a snapshot: the sum of every blob's size in its committed tree. */
-    private long treeSize(String repoRoot, String sha) {
-        String out = exec("git", "-C", repoRoot, "ls-tree", "-r", "-l", sha);
-        long total = 0;
-        for (String line : out.split("\\R")) {
-            if (line.isBlank()) {
-                continue;
-            }
-            // ls-tree -l row: "<mode> <type> <object> <size>\t<path>"; the size is only numeric for
-            // blobs (trees read "-"), so gate on the type token and take the 4th whitespace field.
-            String[] parts = line.split("\\s+");
-            if (parts.length >= 4 && "blob".equals(parts[1])) {
-                try {
-                    total += Long.parseLong(parts[3]);
-                } catch (NumberFormatException ignored) {
-                    // a '-' size (or any anomaly) contributes nothing to the rough figure
-                }
-            }
-        }
-        return total;
-    }
-
-    /** Resolve the directory that will hold per-session worktree checkouts. */
-    private Path resolveRoot(String repoRoot) {
-        if (configuredRoot != null && !configuredRoot.isBlank()) {
-            return Path.of(configuredRoot).toAbsolutePath().normalize();
-        }
-        Path repo = Path.of(repoRoot).toAbsolutePath().normalize();
-        return repo.resolveSibling(".bridged-worktrees");
-    }
-
-    private String nonce() {
-        return String.format("%06x", random.nextInt(1 << 24)) + "-" + seq.incrementAndGet();
-    }
-
-    private boolean isTracked(Path worktreeRoot, String rel) {
-        return exitCode("git", "-C", worktreeRoot.toString(), "ls-files", "--error-unmatch", rel) == 0;
-    }
-
-    /**
-     * Run a command and return its stdout. Non-zero exit → {@link WorktreeException} with both
-     * stdout and stderr (merged by redirectErrorStream).
-     */
-    private String exec(String... command) {
-        return exec(Map.of(), command);
-    }
-
-    /** Same as {@link #exec(String...)}, with extra environment variables set on the child process. */
-    private String exec(Map<String, String> extraEnv, String... command) {
-        String out;
-        int code;
-        Process p;
-        try {
-            ProcessBuilder pb = new ProcessBuilder(command).redirectErrorStream(true);
-            if (extraEnv != null && !extraEnv.isEmpty()) {
-                pb.environment().putAll(extraEnv);
-            }
-            p = pb.start();
-        } catch (IOException e) {
-            throw new WorktreeException("failed to start " + command[0] + ": " + e.getMessage(), e);
-        }
-        try (BufferedReader r = new BufferedReader(new InputStreamReader(p.getInputStream(), StandardCharsets.UTF_8))) {
-            out = r.lines().collect(Collectors.joining("\n"));
-        } catch (IOException e) {
-            p.destroyForcibly();
-            throw new UncheckedIOException(e);
-        }
-        try {
-            if (!p.waitFor(30, TimeUnit.SECONDS)) {
-                p.destroyForcibly();
-                throw new WorktreeException("command timed out: " + String.join(" ", command) + "\n" + out);
-            }
-            code = p.exitValue();
-        } catch (InterruptedException e) {
-            p.destroyForcibly();
-            Thread.currentThread().interrupt();
-            throw new WorktreeException("interrupted waiting for command: " + String.join(" ", command), e);
-        }
-        if (code != 0) {
-            throw new WorktreeException("exit " + code + " for: " + String.join(" ", command)
-                    + (out.isBlank() ? "" : "\n" + out));
-        }
-        return out;
-    }
-
-    private int exitCode(String... command) {
-        Process p;
-        try {
-            p = new ProcessBuilder(command).redirectErrorStream(true).start();
-        } catch (IOException e) {
-            throw new WorktreeException("failed to start " + command[0] + ": " + e.getMessage(), e);
-        }
-        try {
-            if (!p.waitFor(30, TimeUnit.SECONDS)) {
-                p.destroyForcibly();
-                throw new WorktreeException("command timed out: " + String.join(" ", command));
-            }
-            return p.exitValue();
-        } catch (InterruptedException e) {
-            p.destroyForcibly();
-            Thread.currentThread().interrupt();
-            throw new WorktreeException("interrupted waiting for command: " + String.join(" ", command), e);
-        }
-    }
-}

bridged/src/test/java/dev/ltms/bridged/RequiredSecretEnvVarsTest.java

@@ -1,114 +0,0 @@
-package dev.ltms.bridged;
-
-import dev.ltms.bridged.config.BridgedConfig;
-import org.junit.jupiter.api.Test;
-import org.junit.jupiter.api.io.TempDir;
-
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.List;
-import java.util.Map;
-
-import static org.junit.jupiter.api.Assertions.assertEquals;
-import static org.junit.jupiter.api.Assertions.assertFalse;
-import static org.junit.jupiter.api.Assertions.assertTrue;
-
-/**
- * CB-594: {@link Bridged#requiredSecretEnvVars(BridgedConfig)} is what decides what the startup
- * secret report checks — it must derive that set from the config, not a hand-written list, or a
- * new profile's token silently stops being reported.
- */
-class RequiredSecretEnvVarsTest {
-
-    private static BridgedConfig load(Path dir, String yaml) throws Exception {
-        Path f = dir.resolve("bridged.yaml");
-        Files.writeString(f, yaml);
-        return BridgedConfig.load(f);
-    }
-
-    @Test
-    void collectsATokenEnvPerNonSubscriptionProfile(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, """
-                profiles:
-                  local:
-                    baseUrl: http://gx00.gw:8000
-                    tokenEnv: AI_GATEWAY_TOKEN
-                """);
-
-        Map<String, List<String>> required = Bridged.requiredSecretEnvVars(cfg);
-
-        assertTrue(required.containsKey("AI_GATEWAY_TOKEN"));
-        assertEquals(List.of("profile 'local' tokenEnv"), required.get("AI_GATEWAY_TOKEN"));
-    }
-
-    @Test
-    void aSubscriptionProfileNeedsNoTokenEnv(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, """
-                profiles:
-                  opus:
-                    subscription: true
-                    model: claude-opus-5
-                """);
-
-        assertTrue(Bridged.requiredSecretEnvVars(cfg).isEmpty(),
-                "subscription: true never reads ANTHROPIC_AUTH_TOKEN — see Profile#isSubscription");
-    }
-
-    @Test
-    void gitTokenEnvIsOptInAndCollectedWhenSet(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, """
-                profiles:
-                  local:
-                    baseUrl: http://gx00.gw:8000
-                    tokenEnv: AI_GATEWAY_TOKEN
-                    gitTokenEnv: WORKER_GITEA_TOKEN
-                """);
-
-        Map<String, List<String>> required = Bridged.requiredSecretEnvVars(cfg);
-
-        assertTrue(required.containsKey("WORKER_GITEA_TOKEN"));
-        assertEquals(List.of("profile 'local' gitTokenEnv"), required.get("WORKER_GITEA_TOKEN"));
-    }
-
-    @Test
-    void noGitTokenEnvMeansNothingIsRequiredForIt(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, """
-                profiles:
-                  local:
-                    baseUrl: http://gx00.gw:8000
-                    tokenEnv: AI_GATEWAY_TOKEN
-                """);
-
-        assertFalse(Bridged.requiredSecretEnvVars(cfg).containsKey("WORKER_GITEA_TOKEN"));
-    }
-
-    @Test
-    void aVarSharedByTwoProfilesIsReportedOnceNamingBoth(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, """
-                profiles:
-                  local:
-                    baseUrl: http://gx00.gw:8000
-                    tokenEnv: AI_GATEWAY_TOKEN
-                    gitTokenEnv: WORKER_GITEA_TOKEN
-                  gx:
-                    kind: opencode
-                    baseUrl: https://llm.ltms.dev/v1
-                    tokenEnv: AI_GATEWAY_TOKEN
-                    gitTokenEnv: WORKER_GITEA_TOKEN
-                """);
-
-        Map<String, List<String>> required = Bridged.requiredSecretEnvVars(cfg);
-
-        assertEquals(List.of("profile 'local' tokenEnv", "profile 'gx' tokenEnv"),
-                required.get("AI_GATEWAY_TOKEN"));
-        assertEquals(List.of("profile 'local' gitTokenEnv", "profile 'gx' gitTokenEnv"),
-                required.get("WORKER_GITEA_TOKEN"));
-    }
-
-    @Test
-    void noProfilesMeansNothingIsRequired(@TempDir Path dir) throws Exception {
-        BridgedConfig cfg = load(dir, "bind:\n  host: 127.0.0.1\n  port: 8765\n");
-
-        assertTrue(Bridged.requiredSecretEnvVars(cfg).isEmpty());
-    }
-}

bridged/src/test/java/dev/ltms/bridged/health/FleetHealthMonitorTest.java

@@ -1,170 +0,0 @@
-package dev.ltms.bridged.health;
-
-import ch.qos.logback.classic.Logger;
-import ch.qos.logback.classic.spi.ILoggingEvent;
-import ch.qos.logback.core.read.ListAppender;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.FakeHerdr;
-import dev.ltms.bridged.inject.Injector;
-import dev.ltms.bridged.msg.InMemoryReplyInbox;
-import dev.ltms.bridged.msg.MessageService;
-import dev.ltms.bridged.msg.Rendezvous;
-import dev.ltms.bridged.session.SessionManager;
-import dev.ltms.bridged.member.ClaudeCodeLauncher;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.guard.SubscriptionGuard;
-import dev.ltms.bridged.config.BridgedConfig;
-import org.junit.jupiter.api.Test;
-import org.slf4j.LoggerFactory;
-
-import java.util.Map;
-import java.util.Set;
-import java.util.concurrent.Executors;
-import java.util.function.BiConsumer;
-
-import static org.junit.jupiter.api.Assertions.assertEquals;
-import static org.junit.jupiter.api.Assertions.assertTrue;
-
-class FleetHealthMonitorTest {
-    @Test void oneTickUsesOneFleetListForAnyRosterSize() {
-        FakeHerdr herdr = new FakeHerdr();
-        AgentControl agents = new AgentControl(herdr);
-        BridgedConfig.Profile profile = new BridgedConfig.Profile("test", "http://test:1", null,
-                null, null, null, null, null, null, null, null, null);
-        ClaudeCodeLauncher launcher = new ClaudeCodeLauncher(agents, new WorkspaceControl(herdr),
-                new SubscriptionGuard(Set.of("test")), Map.of("test", profile), "test", _ -> "token");
-        SessionManager sessions = new SessionManager(launcher);
-        sessions.acquire("test", null, null, null);
-        sessions.acquire("test", null, null, null);
-        herdr.calls.clear();
-        MessageService messages = new MessageService(agents, new Injector(agents), new Rendezvous(), new InMemoryReplyInbox());
-        var scheduler = Executors.newSingleThreadScheduledExecutor();
-        FleetHealthMonitor monitor = new FleetHealthMonitor(agents, sessions::roster, messages, scheduler, () -> 1, 60,
-                (_, _) -> { });
-        monitor.tick();
-        monitor.stop();
-        assertEquals(1, herdr.calls.stream().filter(call -> call.method().equals("agent.list")).count());
-    }
-
-    @Test void failedTickDoesNotStopTheNextTick() {
-        FakeHerdr herdr = new FakeHerdr().healthy(false);
-        AgentControl agents = new AgentControl(herdr);
-        var scheduler = Executors.newSingleThreadScheduledExecutor();
-        FleetHealthMonitor monitor = new FleetHealthMonitor(agents, java.util.List::of,
-                new MessageService(agents, new Injector(agents), new Rendezvous(), new InMemoryReplyInbox()),
-                scheduler, () -> 1, 60, (_, _) -> { });
-        monitor.tick();
-        herdr.healthy(true);
-        monitor.tick();
-        monitor.stop();
-        assertEquals(2, herdr.calls.stream().filter(call -> call.method().equals("agent.list")).count());
-    }
-
-    @Test void faultTransitionLogsOnlyOnceUntilItChanges() {
-        Logger logger = (Logger) LoggerFactory.getLogger(FleetHealthMonitor.class);
-        ListAppender<ILoggingEvent> appender = new ListAppender<>();
-        appender.start();
-        logger.addAppender(appender);
-        try {
-            FakeHerdr herdr = new FakeHerdr();
-            AgentControl agents = new AgentControl(herdr);
-            var scheduler = Executors.newSingleThreadScheduledExecutor();
-            FleetHealthMonitor monitor = new FleetHealthMonitor(agents, java.util.List::of,
-                    new MessageService(agents, new Injector(agents), new Rendezvous(), new InMemoryReplyInbox()),
-                    scheduler, () -> 1, 60, (_, _) -> { });
-            monitor.reportTransition("term_a", HealthState.TURN_BOUNDARY_LOST);
-            monitor.reportTransition("term_a", HealthState.TURN_BOUNDARY_LOST);
-            monitor.stop();
-            assertEquals(1, appender.list.stream().filter(event -> event.getFormattedMessage()
-                    .contains("member=term_a state=TURN_BOUNDARY_LOST")).count());
-        } finally {
-            logger.detachAppender(appender);
-        }
-    }
-
-    // --- CB-580: a member that reaches GONE/NEVER_READY must fail its waiting tickets
-
-    private static FleetHealthMonitor monitorWith(BiConsumer<String, String> failTarget) {
-        FakeHerdr herdr = new FakeHerdr();
-        AgentControl agents = new AgentControl(herdr);
-        var scheduler = Executors.newSingleThreadScheduledExecutor();
-        return new FleetHealthMonitor(agents, java.util.List::of,
-                new MessageService(agents, new Injector(agents), new Rendezvous(), new InMemoryReplyInbox()),
-                scheduler, () -> 1, 60, failTarget);
-    }
-
-    @Test void terminalTransitionFailsTheTargetOnce() {
-        RecordingFailTarget failTarget = new RecordingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.stop();
-        assertEquals(1, failTarget.calls.size());
-        assertEquals("term_a", failTarget.calls.get(0).target());
-        assertTrue(failTarget.calls.get(0).reason().contains("GONE"));
-    }
-
-    @Test void neverReadyNamesItselfAsTheReason() {
-        RecordingFailTarget failTarget = new RecordingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.NEVER_READY);
-        monitor.stop();
-        assertEquals(1, failTarget.calls.size());
-        assertTrue(failTarget.calls.get(0).reason().contains("NEVER_READY"));
-    }
-
-    @Test void stayingInATerminalStateProducesOneFailureNotN() {
-        RecordingFailTarget failTarget = new RecordingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.stop();
-        assertEquals(1, failTarget.calls.size());
-    }
-
-    @Test void aNonTerminalFaultStateDoesNotFailTheTarget() {
-        RecordingFailTarget failTarget = new RecordingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.TURN_BOUNDARY_LOST);
-        monitor.stop();
-        assertEquals(0, failTarget.calls.size());
-    }
-
-    @Test void failTargetRetryIsBounded() {
-        AlwaysThrowingFailTarget failTarget = new AlwaysThrowingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.stop();
-        assertEquals(FleetHealthMonitor.MAX_FAIL_TARGET_ATTEMPTS, failTarget.calls);
-    }
-
-    @Test void exhaustedRetryStillDoesNotRefireOnAnUnchangedTick() {
-        AlwaysThrowingFailTarget failTarget = new AlwaysThrowingFailTarget();
-        FleetHealthMonitor monitor = monitorWith(failTarget);
-        monitor.reportTransition("term_a", HealthState.GONE);
-        int afterFirstTransition = failTarget.calls;
-        monitor.reportTransition("term_a", HealthState.GONE);
-        monitor.stop();
-        assertEquals(afterFirstTransition, failTarget.calls);
-    }
-
-    private record RecordedCall(String target, String reason) { }
-
-    private static final class RecordingFailTarget implements BiConsumer<String, String> {
-        final java.util.List<RecordedCall> calls = new java.util.ArrayList<>();
-
-        @Override public void accept(String target, String reason) {
-            calls.add(new RecordedCall(target, reason));
-        }
-    }
-
-    private static final class AlwaysThrowingFailTarget implements BiConsumer<String, String> {
-        int calls = 0;
-
-        @Override public void accept(String target, String reason) {
-            calls++;
-            throw new RuntimeException("boom");
-        }
-    }
-}

bridged/src/test/java/dev/ltms/bridged/herdr/PaneLocatorTest.java

@@ -1,27 +0,0 @@
-package dev.ltms.bridged.herdr;
-
-import org.junit.jupiter.api.Test;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-/** Unit tests for PID → pane resolution (the herdr half of connection-based MCP identity). */
-class PaneLocatorTest {
-
-    private final PaneLocator loc = new PaneLocator(new FakeHerdr());
-
-    @Test
-    void resolvesTerminalForAForegroundPid() {
-        assertEquals("term_a", loc.terminalForPid(FakeHerdr.WORKER_PID));
-    }
-
-    @Test
-    void nullForAPidInNoPane() {
-        assertNull(loc.terminalForPid(999_999));
-    }
-
-    @Test
-    void nullForNonPositivePid() {
-        assertNull(loc.terminalForPid(0));
-        assertNull(loc.terminalForPid(-1));
-    }
-}

bridged/src/test/java/dev/ltms/bridged/inject/CompletionResolverTest.java

@@ -1,501 +0,0 @@
-package dev.ltms.bridged.inject;
-
-import ch.qos.logback.classic.Level;
-import ch.qos.logback.classic.LoggerContext;
-import ch.qos.logback.classic.spi.ILoggingEvent;
-import ch.qos.logback.core.read.ListAppender;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.FakeHerdr;
-import dev.ltms.bridged.msg.Rendezvous;
-import dev.ltms.bridged.msg.TestTurnTokens;
-import dev.ltms.bridged.msg.TurnToken;
-import org.junit.jupiter.api.Test;
-import org.slf4j.LoggerFactory;
-
-import java.util.Set;
-import java.util.regex.Pattern;
-
-import static org.junit.jupiter.api.Assertions.assertEquals;
-import static org.junit.jupiter.api.Assertions.assertFalse;
-import static org.junit.jupiter.api.Assertions.assertTrue;
-
-/** Unit behaviour of the CB-106 completion resolver in isolation from the injector. */
-class CompletionResolverTest {
-
-    @Test
-    void skipsTheScrapeWhenNoSendIsWaiting() {
-        FakeHerdr herdr = new FakeHerdr();
-        Rendezvous rendezvous = new Rendezvous(); // no waiter opened
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        resolver.resolve("term_a", null); // no in-flight turn captured for this target
-
-        assertFalse(herdr.called("agent.read"),
-                "a turn nobody is blocked on must not cost a transcript scrape");
-    }
-
-    @Test
-    void failSkipsTheScrapeWhenNoSendIsWaiting() {
-        FakeHerdr herdr = new FakeHerdr();
-        Rendezvous rendezvous = new Rendezvous(); // no waiter opened
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        resolver.fail("term_a", null); // no in-flight turn, and no registered waiter to fall back to
-
-        assertFalse(herdr.called("agent.read"),
-                "a wedge nobody is blocked on must not cost a transcript scrape");
-    }
-
-    @Test
-    void captureBaselineSkipsTheReadWhenNoSendIsWaiting() {
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ X\n❯ ");
-        Rendezvous rendezvous = new Rendezvous(); // no waiter opened
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        resolver.captureBaseline("term_a", TestTurnTokens.inert("term_a")); // no send to attribute a later completion to
-
-        assertFalse(herdr.called("agent.read"),
-                "with no waiting send there is no turn to baseline — skip the scrape");
-    }
-
-    // --- CB-115 clean scrape: extract the last assistant block ----------------
-
-    @Test
-    void extractsTheLastAssistantBlockStrippingChrome() {
-        String raw = """
-                ⏺ Reading the file…
-
-                ⏺ Done. The bug was an off-by-one in the loop bound.
-
-                ╭──────────────────────────────────────╮
-                │ >                                    │
-                ╰──────────────────────────────────────╯
-                  ⏵⏵ auto mode on · ? for shortcuts
-                """;
-        assertEquals("Done. The bug was an off-by-one in the loop bound.",
-                CompletionResolver.lastAssistantBlock(raw));
-    }
-
-    @Test
-    void keepsMultiLineAssistantContent() {
-        String raw = "⏺ Line one.\nLine two.\n❯ ";
-        assertEquals("Line one.\nLine two.", CompletionResolver.lastAssistantBlock(raw));
-    }
-
-    @Test
-    void fallsBackToRawTextWhenThereIsNoMarker() {
-        String raw = "plain worker output with no glyph";
-        assertEquals("plain worker output with no glyph", CompletionResolver.lastAssistantBlock(raw));
-    }
-
-    @Test
-    void blankScrapeYieldsEmpty() {
-        assertTrue(CompletionResolver.lastAssistantBlock("").isEmpty());
-        assertTrue(CompletionResolver.lastAssistantBlock(null).isEmpty());
-    }
-
-    @Test
-    void stripsSpinnerAndRuleChrome() {
-        String raw = """
-                ⏺ Channel check confirmed — your message got through.
-
-                ✻ Brewed for 11s
-
-                ─────────────────────────────────────
-                """;
-        assertEquals("Channel check confirmed — your message got through.",
-                CompletionResolver.lastAssistantBlock(raw));
-    }
-
-    @Test
-    void cutsANextTurnPromptEchoAndTrailingTipsFromTheBlock() {
-        // The exact turn-2 leak: the scrape captured the settled answer, then a "✻ Cooked" spinner,
-        // then the NEXT turn's echoed prompt, then a "✶ Forming…" spinner and trailing tips/warnings
-        // whose lines (⎿, ⚠) are not themselves chrome-terminated. Stopping at the first boundary
-        // (the ✻ spinner) is what keeps every one of those interface lines out of the reply.
-        String raw = """
-                ⏺ Channel confirmed — the bridge reply delivered successfully.
-
-                ✻ Cooked for 9s
-
-                ❯ Thanks. Now a small task: what is 17 * 23? Show just the number.
-
-
-
-                ✶ Forming…
-                  ⎿  Tip: Name your conversations with /rename
-                  ⚠ claude.ai connectors are disabled because ANTHROPIC_API_KEY is set
-                """;
-        assertEquals("Channel confirmed — the bridge reply delivered successfully.",
-                CompletionResolver.lastAssistantBlock(raw));
-    }
-
-    // --- CB-115 misattribution guard: suppress a stale (unchanged) completion -------
-
-    @Test
-    void suppressesACompletionWhoseScrapeIsUnchangedFromDelivery() {
-        // Rapid back-to-back turn: the pane still shows the PREVIOUS turn's answer when this turn's
-        // (misattributed) completion boundary fires. The scrape == the delivery baseline, so the
-        // send must NOT be resolved with the stale answer.
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ 391\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");   // a send is blocked on this turn
-        // The turn as captured at delivery: its waiter, and the previous turn's answer still on screen.
-        var turn = new CompletionResolver.InFlight(waiter, "391");
-        resolver.resolve("term_a", turn);          // scrape still "391" == baseline → suppress
-
-        assertFalse(waiter.isDone(), "a completion with no output change must not resolve the send");
-        assertTrue(rendezvous.isWaiting("term_a"), "the send stays waiting for a real reply");
-    }
-
-    @Test
-    void resolvesACompletionWhoseScrapeChangedSinceDelivery() {
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ No, 391 = 17 × 23.\n❯ "); // the worker's real answer
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        // Delivery baseline was the previous turn's "391"; the scrape now differs → resolve.
-        var turn = new CompletionResolver.InFlight(waiter, "391");
-        resolver.resolve("term_a", turn);
-
-        assertTrue(waiter.isDone(), "a completion with new output must resolve the send");
-        assertEquals(Rendezvous.Kind.COMPLETION, waiter.getNow(null).kind());
-        assertEquals("No, 391 = 17 × 23.", waiter.getNow(null).text());
-    }
-
-    @Test
-    void marksAClippedCompletionPaneTail() {
-        String block = "⏺ " + "x".repeat(CompletionResolver.MAX_SCRAPE_CHARS + 1) + "\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals("x".repeat(CompletionResolver.MAX_SCRAPE_CHARS)
-                        + "\n[Pane tail clipped: member did not call bridge_reply.]",
-                waiter.getNow(null).text());
-    }
-
-    @Test
-    void leavesAnUnclippedCompletionPaneTailUnmarked() {
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ complete report\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals("complete report", waiter.getNow(null).text());
-    }
-
-    @Test
-    void resolvesSynchronouslyBeforePostTurnContextClearing() {
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ previous answer\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-        var waiter = rendezvous.open("term_a");
-        resolver.captureBaseline("term_a", new TurnToken("term_a", waiter));
-        herdr.readText("⏺ answer that /clear would erase\n❯ ");
-
-        resolver.resolveBeforePostAction("term_a");
-
-        assertTrue(waiter.isDone(), "the answer is captured before the adapter sends /clear");
-        assertEquals("answer that /clear would erase", waiter.getNow(null).text());
-    }
-
-    @Test
-    void suppressesAnUnchangedCompletionEvenWhenTheBlockExceedsTheScrapeCap() {
-        // The fan-out issue-hunt finding: captureBaseline once stored the RAW (unclipped) assistant
-        // block while resolve compares against a clip()'d tail. For a block longer than MAX_SCRAPE_CHARS
-        // the two capped representations differ even when the pane never changed, so the CB-115
-        // byte-identical guard failed to fire and a stale completion could resolve the send. Both sides
-        // must clip identically. The returned-text marker is added only after this comparison, so an
-        // unchanged >cap block on rapid back-to-back turns still stays suppressed.
-        String longBlock = "⏺ " + "x".repeat(CompletionResolver.MAX_SCRAPE_CHARS + 500) + "\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(longBlock);
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");        // a send is blocked on this turn
-        resolver.captureBaseline("term_a", new TurnToken("term_a", waiter)); // baseline is the clipped >cap block
-        var turn = resolver.inFlight("term_a");
-        assertEquals(CompletionResolver.MAX_SCRAPE_CHARS, turn.baseline().length(),
-                "the delivery baseline is clipped to the same cap resolve() applies to the tail");
-
-        resolver.resolve("term_a", turn);              // scrape unchanged → clipped tail == baseline → suppress
-
-        assertFalse(waiter.isDone(),
-                "an unchanged >cap block must still be recognised as stale and suppressed");
-        assertTrue(rendezvous.isWaiting("term_a"), "the send stays waiting for a real reply");
-    }
-
-    @Test
-    void resolvesWhenThereIsNoBaseline() {
-        // No delivery baseline (e.g. the pre-turn read failed) ⇒ never suppress; the completion resolves.
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ hello\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertTrue(waiter.isDone(), "with no baseline a completion resolves as before");
-        assertEquals("hello", waiter.getNow(null).text());
-    }
-
-    @Test
-    void resolvesWhenTheScrapeItselfFailsEvenWithABaselinePresent() {
-        // The most important branch of the CB-115 guard: a failed read means the resolver could not
-        // SEE the screen — "couldn't see", not "no change". It must still resolve the send (an empty
-        // tail beats hanging until the caller's timeout), even though a baseline was captured. The
-        // baseline here is "" (an empty pane at delivery), so without the !scrapeFailed clause the
-        // byte-identical guard would wrongly match the empty tail and suppress.
-        FakeHerdr herdr = new FakeHerdr().healthy(false); // agent.read throws HerdrException
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        var turn = new CompletionResolver.InFlight(waiter, ""); // empty pane baselined at delivery
-        resolver.resolve("term_a", turn);
-
-        assertTrue(waiter.isDone(),
-                "a failed scrape must still resolve the send, not hang until the caller's timeout");
-        assertEquals(Rendezvous.Kind.COMPLETION, waiter.getNow(null).kind());
-        assertEquals("", waiter.getNow(null).text(), "the tail is empty because the screen was unreadable");
-    }
-
-    // --- CB-115/CB-116 fail guard: an already-done or absent waiter is left alone ---------
-
-    @Test
-    void failLeavesAnAlreadyResolvedWaiterUntouchedAndSkipsTheScrape() {
-        // The send was already resolved (e.g. by the worker's explicit reply) before fail fired.
-        // fail must not overwrite that value, and must not even scrape the worker — nobody needs it.
-        FakeHerdr herdr = new FakeHerdr().readText("an error screen");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        var turn = new CompletionResolver.InFlight(waiter, null);
-        assertTrue(rendezvous.resolveCompletion(waiter, "already replied"));
-
-        resolver.fail("term_a", turn);
-
-        assertFalse(herdr.called("agent.read"),
-                "fail must not scrape a waiter that is already done");
-        assertEquals(Rendezvous.Kind.COMPLETION, waiter.getNow(null).kind(),
-                "fail must not overwrite the existing resolution");
-        assertEquals("already replied", waiter.getNow(null).text());
-    }
-
-    @Test
-    void failFallsBackToTheRegisteredWaiterWhenThereIsNoInFlightTurn() {
-        // A never-delivered readiness failure has no in-flight record but still has a blocked send;
-        // fail falls back to the waiter currently registered on the Rendezvous and fails it.
-        FakeHerdr herdr = new FakeHerdr().readText("stuck on an error screen");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a"); // send registered, but no captureBaseline ever ran
-        resolver.fail("term_a", null);          // no in-flight turn → fall back to the registered waiter
-
-        assertTrue(waiter.isDone(), "fail falls back to the registered waiter when no turn is in flight");
-        assertEquals(Rendezvous.Kind.FAILED, waiter.getNow(null).kind());
-        assertEquals("stuck on an error screen", waiter.getNow(null).text());
-    }
-
-    @Test
-    void failIsLoggedAtWarnWithTheReason() {
-        // CB-564: this used to be a bare DEBUG "failed send to X via turn-stall fallback" — a symptom
-        // with no cause, and below the level anyone watching for member health would see. A fail that
-        // resolves a caller's blocked send is at least WARN and must carry the reason.
-        LoggerContext ctx = (LoggerContext) LoggerFactory.getILoggerFactory();
-        ch.qos.logback.classic.Logger resolverLog =
-                (ch.qos.logback.classic.Logger) LoggerFactory.getLogger(CompletionResolver.class);
-        ListAppender<ILoggingEvent> appender = new ListAppender<>();
-        appender.setContext(ctx);
-        appender.start();
-        resolverLog.addAppender(appender);
-        resolverLog.setLevel(Level.WARN);
-        try {
-            FakeHerdr herdr = new FakeHerdr().readText("stuck on an error screen");
-            Rendezvous rendezvous = new Rendezvous();
-            CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-            var waiter = rendezvous.open("term_a");
-
-            resolver.fail("term_a", null);
-
-            String warn = appender.list.stream()
-                    .filter(e -> e.getLevel().equals(Level.WARN))
-                    .map(ILoggingEvent::getFormattedMessage)
-                    .findFirst()
-                    .orElse("no turn-stall WARN logged");
-            assertTrue(warn.contains("term_a"), "the log names the target: " + warn);
-            assertTrue(warn.contains("stuck on an error screen"), "the log carries the reason: " + warn);
-            assertTrue(waiter.isDone());
-        } finally {
-            resolverLog.detachAppender(appender);
-        }
-    }
-
-    // --- CB-116 waiter identity: a late completion never crosses into the next turn ---------
-
-    @Test
-    void aLateCompletionForOneTurnNeverResolvesTheNextTurnsWaiter() {
-        // The cross-turn stale reply the conversation test surfaced: turn N's completion fallback
-        // fires AFTER turn N was resolved by an explicit bridge_reply and turn N+1 has opened its own
-        // waiter on the same session. Resolving "whatever is waiting now" would hand turn N's stale
-        // scrape to turn N+1; targeting turn N's captured waiter makes the late completion a no-op.
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ turn N answer\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiterN = rendezvous.open("term_a"); // turn N's send
-        // The turn as the injector captured it at delivery (waiter + pre-turn baseline).
-        var turnN = new CompletionResolver.InFlight(waiterN, "an earlier answer");
-
-        // Turn N is resolved by the worker's explicit reply, and its send deregisters the waiter.
-        assertTrue(rendezvous.resolve("term_a", "N replied"));
-        rendezvous.close("term_a", waiterN); // the sender's finally, before the next turn opens
-
-        // Turn N+1's send opens its own waiter on the same session (CB-548: open fails if the
-        // previous waiter is still registered, so a clean turn deregisters it first as above).
-        var waiterN1 = rendezvous.open("term_a");
-
-        resolver.resolve("term_a", turnN); // turn N's completion fallback finally fires
-
-        assertFalse(waiterN1.isDone(), "turn N's late completion must not resolve turn N+1's waiter");
-        assertEquals(Rendezvous.Kind.REPLY, waiterN.getNow(null).kind(),
-                "turn N stays resolved by its own reply");
-        assertTrue(rendezvous.isWaiting("term_a"), "turn N+1 is still awaiting its own resolution");
-    }
-
-    // --- CB-578 stage A: backend-exhausted classification ---------------------------------
-
-    @Test
-    void classifiesAMatchingScrapeAsBackendExhaustedInsteadOfACompletedReply() {
-        String block = "⏺ Working on it...\nThe usage limit has been reached. Try again later.\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        ExhaustedPatternLookup patterns = target -> Pattern.compile("usage limit has been reached");
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, patterns, ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertTrue(waiter.isDone(), "a matching scrape still resolves the blocked send");
-        assertEquals(Rendezvous.Kind.BACKEND_EXHAUSTED, waiter.getNow(null).kind(),
-                "not reported as a completed reply — the classification is distinct");
-    }
-
-    @Test
-    void theExhaustedReasonCarriesTheMatchedLine() {
-        String block = "⏺ Working on it...\nThe usage limit has been reached. Try again later.\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        ExhaustedPatternLookup patterns = target -> Pattern.compile("usage limit has been reached");
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, patterns, ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals("backend exhausted (usage limit): The usage limit has been reached. Try again later.",
-                waiter.getNow(null).text(), "the reason names the real cause and carries the matched line");
-    }
-
-    @Test
-    void aWinningBackendExhaustedClassificationNotifiesTheExhaustionSink() {
-        String block = "⏺ Working on it...\nThe usage limit has been reached. Try again later.\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        ExhaustedPatternLookup patterns = target -> Pattern.compile("usage limit has been reached");
-        java.util.List<String> notified = new java.util.ArrayList<>();
-        ExhaustionSink sink = (target, reason) -> notified.add(target + ": " + reason);
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, patterns, sink);
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals(1, notified.size(), "the sink is notified exactly once for the winning classification");
-        assertTrue(notified.get(0).startsWith("term_a: "), "the sink is told which target exhausted");
-        assertTrue(notified.get(0).contains("The usage limit has been reached"),
-                "the sink is told the matched reason: " + notified.get(0));
-    }
-
-    @Test
-    void aLosingBackendExhaustedClassificationNeverNotifiesTheExhaustionSink() {
-        // The waiter was already resolved (e.g. by the worker's own reply) before this scrape landed —
-        // resolveExhausted loses the race and must return false, so the sink must not fire either.
-        String block = "⏺ Working on it...\nThe usage limit has been reached. Try again later.\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        ExhaustedPatternLookup patterns = target -> Pattern.compile("usage limit has been reached");
-        java.util.List<String> notified = new java.util.ArrayList<>();
-        ExhaustionSink sink = (target, reason) -> notified.add(target + ": " + reason);
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, patterns, sink);
-
-        var waiter = rendezvous.open("term_a");
-        var turn = new CompletionResolver.InFlight(waiter, null);
-        assertTrue(rendezvous.resolveCompletion(waiter, "already replied"));
-
-        resolver.resolve("term_a", turn);
-
-        assertTrue(notified.isEmpty(), "a classification that loses the race must not quarantine anything");
-        assertEquals("already replied", waiter.getNow(null).text(), "the earlier resolution stands untouched");
-    }
-
-    @Test
-    void aNonMatchingScrapeResolvesAsAnOrdinaryCompletion() {
-        FakeHerdr herdr = new FakeHerdr().readText("⏺ complete report\n❯ ");
-        Rendezvous rendezvous = new Rendezvous();
-        ExhaustedPatternLookup patterns = target -> Pattern.compile("usage limit has been reached");
-        CompletionResolver resolver = new CompletionResolver(new AgentControl(herdr), rendezvous, patterns, ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals(Rendezvous.Kind.COMPLETION, waiter.getNow(null).kind(),
-                "a scrape that does not match the pattern is an ordinary completion");
-        assertEquals("complete report", waiter.getNow(null).text());
-    }
-
-    @Test
-    void aProfileWithNoConfiguredPatternKeepsTodaysCompletionFallbackUnchanged() {
-        // Even a scrape that WOULD have matched some other profile's pattern must resolve as a
-        // plain completion when this target's own profile has none configured (CB-578 criterion 4).
-        String block = "⏺ The usage limit has been reached.\n❯ ";
-        FakeHerdr herdr = new FakeHerdr().readText(block);
-        Rendezvous rendezvous = new Rendezvous();
-        CompletionResolver resolver =
-                new CompletionResolver(new AgentControl(herdr), rendezvous, ExhaustedPatternLookup.none(), ExhaustionSink.none());
-
-        var waiter = rendezvous.open("term_a");
-        resolver.resolve("term_a", new CompletionResolver.InFlight(waiter, null));
-
-        assertEquals(Rendezvous.Kind.COMPLETION, waiter.getNow(null).kind(),
-                "no pattern configured for this target's profile ⇒ unchanged completion-fallback behaviour");
-        assertEquals("The usage limit has been reached.", waiter.getNow(null).text());
-    }
-
-    @Test
-    void coverageIsOffWhenNoProfileHasAPatternConfigured() {
-        assertEquals("off (no profile has an exhaustedPattern configured; profiles: [terra])",
-                CompletionResolver.coverage(Set.of("terra"), Set.of()));
-    }
-
-    @Test
-    void coverageIsFullWhenEveryProfileHasAPatternConfigured() {
-        assertEquals("full (all profiles configured: [gx10, terra])",
-                CompletionResolver.coverage(Set.of("terra", "gx10"), Set.of("terra", "gx10")));
-    }
-
-    @Test
-    void coverageIsPartialAndNamesWhichProfilesAreConfigured() {
-        assertEquals("partial (configured: [terra]; not configured: [gx10])",
-                CompletionResolver.coverage(Set.of("terra", "gx10"), Set.of("terra")));
-    }
-}

bridged/src/test/java/dev/ltms/bridged/member/OpenCodeLauncherTest.java

@@ -1,465 +0,0 @@
-package dev.ltms.bridged.member;
-
-import com.fasterxml.jackson.databind.JsonNode;
-import com.fasterxml.jackson.databind.ObjectMapper;
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.FakeHerdr;
-import dev.ltms.bridged.herdr.WorkspaceControl;
-import dev.ltms.bridged.peer.Capability;
-import dev.ltms.bridged.peer.CharterReceipt;
-import dev.ltms.bridged.peer.PeerHandle;
-import dev.ltms.bridged.peer.PeerUnreachableException;
-import dev.ltms.bridged.peer.SpawnRequest;
-import org.junit.jupiter.api.Test;
-import org.junit.jupiter.api.io.TempDir;
-
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.List;
-import java.util.Map;
-import java.util.concurrent.ExecutorService;
-import java.util.concurrent.Executors;
-import java.util.concurrent.Future;
-import java.util.function.Supplier;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-/**
- * The opencode adapter's launch build: a file-based MCP mount + reply-charter instructions (no
- * inline flags, no {@code ANTHROPIC_*}, no guard), the {@code -m} model flag, and the shared base
- * transport (naming, reap, readiness gate) proving the {@link HerdrPeerLauncher} SPI is neutral.
- */
-class OpenCodeLauncherTest {
-
-    private static BridgedConfig.Profile opencodeCfg(String model, String mcpUrl, String gitTokenEnv) {
-        return new BridgedConfig.Profile("gemini", null, model, null, "BRIDGED_WORKER_TOKEN",
-                List.of("opencode"), "tab", "bridged-workers", "opencode: {model} #{n}", mcpUrl,
-                null, null, gitTokenEnv, null, BridgedConfig.Profile.KIND_OPENCODE);
-    }
-
-    /** Gate-disabled launcher whose per-spawn config dirs land under an inspectable temp root. */
-    private static OpenCodeLauncher service(FakeHerdr herdr, Path configRoot, BridgedConfig.Profile cfg) {
-        return new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of(cfg.profile(), cfg), cfg.profile(), k -> "GITEA_ACCESS_TOKEN".equals(k) ? "tok" : null,
-                0, System::currentTimeMillis, () -> { }, configRoot, configRoot);
-    }
-
-    private static OpenCodeLauncher service(FakeHerdr herdr, Path configRoot, BridgedConfig.Profile cfg,
-                                            Supplier<BridgedConfig.Fleet> fleet) {
-        return new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of(cfg.profile(), cfg), cfg.profile(), _ -> null,
-                0, System::currentTimeMillis, () -> { }, configRoot, configRoot, fleet);
-    }
-
-    private static OpenCodeLauncher serviceWithCredentials(FakeHerdr herdr, Path configRoot,
-                                                            BridgedConfig.Profile cfg,
-                                                            BridgedConfig.MemberCredentials creds) {
-        return new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of(cfg.profile(), cfg), cfg.profile(), k -> "GITEA_ACCESS_TOKEN".equals(k) ? "tok" : null,
-                0, System::currentTimeMillis, () -> { }, configRoot, configRoot, null, () -> creds);
-    }
-
-    @SuppressWarnings("unchecked")
-    private static Map<String, Object> lastStart(FakeHerdr herdr) {
-        return (Map<String, Object>) herdr.lastCall("agent.start").params();
-    }
-
-    /** Protocol 19: the worker's env is injected at pane creation (tab.create), not agent.start. */
-    @SuppressWarnings("unchecked")
-    private static Map<String, String> startEnv(FakeHerdr herdr) {
-        Map<String, String> env =
-                (Map<String, String>) ((Map<String, Object>) herdr.lastCall("tab.create").params()).get("env");
-        return env == null ? Map.of() : env;
-    }
-
-    /** Protocol 19: agent.start carries only the args after the kind-resolved executable. */
-    @SuppressWarnings("unchecked")
-    private static List<String> startArgs(FakeHerdr herdr) {
-        return (List<String>) lastStart(herdr).get("args");
-    }
-
-    @Test
-    void writesRemoteMcpConfigAndCharterInstructionsWhenMcpUrlSet(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        BridgedConfig.Fleet fleet = new BridgedConfig.Fleet(Map.of(), Map.of(), Map.of(), Map.of(),
-                Map.of("dev", "role rule"), null);
-        service(herdr, root, opencodeCfg("google/gemini-2.5-pro", "http://127.0.0.1:8765/mcp", null),
-                () -> fleet).spawn();
-
-        Map<String, String> env = startEnv(herdr);
-        assertNull(env.get("ANTHROPIC_BASE_URL"), "opencode carries no ANTHROPIC_* / subscription boundary");
-        String cfgPath = env.get("OPENCODE_CONFIG");
-        assertNotNull(cfgPath, "OPENCODE_CONFIG points the worker at the generated config file");
-        assertTrue(Path.of(cfgPath).startsWith(root), "config file is generated under the injected root");
-
-        // Assert on parsed structure, not substrings: the generated config is real JSON and its
-        // whitespace is the formatter's business, not the contract's.
-        JsonNode json = new ObjectMapper().readTree(Path.of(cfgPath).toFile());
-        assertTrue(json.path("compaction").path("auto").asBoolean(),
-                "spawned opencode peers explicitly enable automatic compaction");
-        JsonNode bridge = json.path("mcp").path("bridge");
-        assertEquals("remote", bridge.path("type").asText(), "bridge is mounted as a remote MCP server");
-        assertEquals("http://127.0.0.1:8765/mcp", bridge.path("url").asText(),
-                "the profile's bridge MCP url is present");
-        assertTrue(bridge.path("enabled").asBoolean(), "the bridge server is enabled");
-        assertTrue(json.path("instructions").isArray() && !json.path("instructions").isEmpty(),
-                "the member charter is mounted via instructions");
-
-        // The instructions entry is a real file path holding the composed member charter.
-        Path charter = Path.of(cfgPath).resolveSibling("member-charter.md");
-        assertTrue(Files.exists(charter), "the charter file the config references was written");
-        assertEquals("role rule\n\n" + HerdrPeerLauncher.REPLY_CHARTER, Files.readString(charter),
-                "the composed charter keeps the role rule first and the reply rule last");
-        assertEquals(charter.toAbsolutePath().toString(), json.path("instructions").get(0).asText(),
-                "instructions names the charter file by its absolute path");
-    }
-
-    @Test
-    void noConfigFileWhenMcpUrlAbsent(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg("google/gemini-2.5-pro", null, null)).spawn();
-
-        assertNull(startEnv(herdr).get("OPENCODE_CONFIG"),
-                "no bridge MCP url → no config file and no OPENCODE_CONFIG");
-    }
-
-    @Test
-    void roleCharterWithoutMcpOrCustomProviderStillWritesAConfig(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        BridgedConfig.Fleet fleet = new BridgedConfig.Fleet(Map.of(), Map.of(), Map.of(), Map.of(),
-                Map.of("dev", "role rule"), null);
-        Path configRoot = Files.createDirectory(root.resolve("configs"));
-        Path checkout = Files.createDirectory(root.resolve("checkout"));
-        service(herdr, configRoot, opencodeCfg("google/gemini-2.5-pro", null, null), () -> fleet).spawn();
-
-        String cfgPath = startEnv(herdr).get("OPENCODE_CONFIG");
-        assertNotNull(cfgPath, "a role charter needs a config even without MCP or custom provider");
-        JsonNode json = new ObjectMapper().readTree(Path.of(cfgPath).toFile());
-        Path charter = Path.of(json.path("instructions").get(0).asText());
-        assertEquals("role rule", Files.readString(charter), "the base-composed role charter is unchanged");
-        assertTrue(json.path("mcp").isMissingNode(), "a charter does not add an MCP mount");
-        assertTrue(charter.startsWith(configRoot), "the charter is written under the temp config root");
-        try (var files = Files.walk(checkout)) {
-            assertFalse(files.anyMatch(path -> path.getFileName().toString().equals("member-charter.md")),
-                    "the worker checkout receives no charter file");
-        }
-    }
-
-    @Test
-    void nullCharterWritesNoCharterFileOrInstructions(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, pinnedCfg("local-vllm/model", "http://127.0.0.1:8000", null),
-                () -> new BridgedConfig.Fleet(Map.of(), Map.of(), Map.of(), Map.of(), Map.of(), null)).spawn();
-
-        String config = startEnv(herdr).get("OPENCODE_CONFIG");
-        assertNotNull(config, "the custom provider still needs a config");
-        JsonNode json = new ObjectMapper().readTree(Path.of(config).toFile());
-        assertTrue(json.path("instructions").isMissingNode(), "a null charter adds no instructions entry");
-        try (var files = Files.walk(root)) {
-            assertFalse(files.anyMatch(path -> path.getFileName().toString().equals("member-charter.md")),
-                    "a null charter creates no charter file");
-        }
-    }
-
-    @Test
-    void concurrentSpawnsWriteSeparateCharterDirectories(@TempDir Path root) throws Exception {
-        BridgedConfig.Profile cfg = opencodeCfg("google/gemini-2.5-pro", "http://127.0.0.1:8765/mcp", null);
-        ExecutorService executor = Executors.newFixedThreadPool(2);
-        try {
-            Future<String> first = executor.submit(() -> spawnConfigPath(root, cfg));
-            Future<String> second = executor.submit(() -> spawnConfigPath(root, cfg));
-
-            Path firstCharter = Path.of(first.get()).resolveSibling("member-charter.md");
-            Path secondCharter = Path.of(second.get()).resolveSibling("member-charter.md");
-            assertNotEquals(firstCharter.getParent(), secondCharter.getParent(),
-                    "each concurrent spawn owns a separate config directory");
-            assertTrue(Files.exists(firstCharter));
-            assertTrue(Files.exists(secondCharter));
-        } finally {
-            executor.shutdownNow();
-        }
-    }
-
-    private static String spawnConfigPath(Path root, BridgedConfig.Profile cfg) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, cfg).spawn();
-        return startEnv(herdr).get("OPENCODE_CONFIG");
-    }
-
-    @Test
-    void passesTheModelAsDashMFlagAlongsideAutoApprove(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg("google/gemini-2.5-pro", null, null)).spawn();
-
-        List<String> args = startArgs(herdr);
-        assertTrue(args.contains("--auto"),
-                "--auto is present alongside -m so a spawned peer never blocks on approval");
-        int m = args.indexOf("-m");
-        assertTrue(m >= 0, "model is selected with -m");
-        assertEquals("google/gemini-2.5-pro", args.get(m + 1), "the provider/model selector follows -m");
-    }
-
-    @Test
-    void autoApproveIsUnconditionalWhenModelBlank(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg(null, null, null)).spawn();
-        assertEquals(List.of("--auto"), startArgs(herdr),
-                "--auto is unconditional: a model-less worker still must never block on approval");
-    }
-
-    @Test
-    void injectsForgeTokenWhenProfileGrantsIt(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg(null, null, "GITEA_ACCESS_TOKEN")).spawn();
-        assertEquals("tok", startEnv(herdr).get("GITEA_TOKEN"),
-                "a git-token profile gets the peer-neutral GITEA_TOKEN grant, same as Claude");
-    }
-
-    /**
-     * CB-596: the config-driven shadow lives in {@link HerdrPeerLauncher#baseEnv}, shared by every
-     * adapter — this pins that the opencode path gets it too, not just Claude's. See the matching
-     * tests in {@code ClaudeCodeLauncherTest} for the full rationale (gitea #82, superseding CB-592's
-     * single hardcoded name).
-     */
-    @Test
-    void aKnownNameNotAllowedIsShadowedWithTheSentinel(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        BridgedConfig.MemberCredentials creds = new BridgedConfig.MemberCredentials(
-                null, List.of("AI_GATEWAY_TOKEN"), List.of("AI_GATEWAY_TOKEN", "GITEA_ACCESS_TOKEN"));
-        serviceWithCredentials(herdr, root, opencodeCfg(null, null, null), creds).spawn();
-
-        String shadowed = startEnv(herdr).get("GITEA_ACCESS_TOKEN");
-        assertNotNull(shadowed, "GITEA_ACCESS_TOKEN must be explicitly overlaid, not left unmentioned");
-        assertFalse(shadowed.isBlank(), "a blank overlay value's override behaviour is unverified — must be non-blank");
-        assertFalse(startEnv(herdr).containsKey("AI_GATEWAY_TOKEN"),
-                "an allow-listed name must get no overlay entry at all");
-    }
-
-    /** No {@code memberCredentials} configured (the pre-CB-596 constructor overloads) blocks nothing. */
-    @Test
-    void noMemberCredentialsConfiguredBlocksNothing(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg(null, null, null)).spawn();
-
-        assertNull(startEnv(herdr).get("GITEA_ACCESS_TOKEN"),
-                "with no memberCredentials configured, nothing is shadowed — config must supply the policy");
-    }
-
-    @Test
-    void capabilitiesDeclareOrphanReapAndMcpAskAndConditionalSelfPr(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        assertEquals(java.util.Set.of(Capability.MID_TURN_ASK, Capability.WORKTREE, Capability.ORPHAN_REAP,
-                        Capability.SESSION_RESUME),
-                service(herdr, root, opencodeCfg(null, null, null)).capabilities(),
-                "opencode can be resumed by its own session id, so SESSION_RESUME is always declared");
-        assertFalse(service(herdr, root, opencodeCfg(null, null, null))
-                        .capabilities().contains(Capability.SESSION_NAME),
-                "opencode has no display-name flag, so SESSION_NAME must NOT be declared");
-        assertTrue(service(herdr, root, opencodeCfg(null, null, "GITEA_ACCESS_TOKEN"))
-                        .capabilities().contains(Capability.SELF_PR),
-                "a git-token profile adds SELF_PR");
-    }
-
-    // --- CB-547: resume + post-hoc session discovery --------------------------------------------
-
-    @Test
-    void aResumeSpawnPassesTheSessionIdAsDashS(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg("google/gemini-2.5-pro", null, null))
-                .spawn(new SpawnRequest(null, null, null, null, "ses_41b79fc90ffeI9E8uZv6VprUn2"));
-
-        List<String> args = startArgs(herdr);
-        int s = args.indexOf("-s");
-        assertTrue(s >= 0, "a resumed spawn carries opencode's -s flag");
-        assertEquals("ses_41b79fc90ffeI9E8uZv6VprUn2", args.get(s + 1),
-                "the resume target id follows -s");
-    }
-
-    @Test
-    void aFreshSpawnCarriesNoSessionFlag(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg(null, null, null))
-                .spawn(new SpawnRequest(null, null, null, null, null));
-
-        assertFalse(startArgs(herdr).contains("-s"),
-                "no resume target → a fresh session with no -s flag");
-    }
-
-    @Test
-    void theHandleDiscoversTheSessionIdForTheWorkersCwdOnlyAfterItAppears(@TempDir Path root,
-                                                                          @TempDir Path discRoot)
-            throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        OpenCodeLauncher launcher = new OpenCodeLauncher(new AgentControl(herdr),
-                new WorkspaceControl(herdr), Map.of("gemini", opencodeCfg(null, null, null)),
-                "gemini", _ -> null, 0, System::currentTimeMillis, () -> { }, root, discRoot);
-
-        PeerHandle handle = launcher.spawn(new SpawnRequest(null, "/work/dir", null));
-
-        // opencode writes the record only when the session is first persisted — the instant the
-        // pane is ready it does not exist, so agentSessionId() is null (never a spawn failure).
-        assertNull(handle.agentSessionId(), "no record yet → null, not a spawn-time block");
-        // Once the record appears (here: same cwd), lazy discovery resolves it — the handle's
-        // session id matches its own worktree, not another's.
-        OpenCodeSessionDiscoveryTest.writeRecord(discRoot, "p1", "ses_a.json",
-                "ses_resolved", "/work/dir", 1000L);
-        assertEquals("ses_resolved", handle.agentSessionId(),
-                "agentSessionId() re-scans and picks up a record that has since been written");
-    }
-
-    @Test
-    void foreignWorkerMatchesOpencodePrefixButNotClaude() {
-        String nonce = "abc123";
-        assertTrue(OpenCodeLauncher.isForeignWorker("opencode-gemini-def456-1", nonce),
-                "an opencode pane from another process is foreign");
-        assertFalse(OpenCodeLauncher.isForeignWorker("opencode-gemini-" + nonce + "-1", nonce),
-                "our own opencode pane (same nonce) is not foreign");
-        assertFalse(OpenCodeLauncher.isForeignWorker("claude-ltms-local-def456-1", nonce),
-                "a claude pane is never reaped by the opencode adapter");
-    }
-
-    @Test
-    void productionConstructorsWireThroughToTheBase() {
-        FakeHerdr herdr = new FakeHerdr();
-        BridgedConfig.Profile cfg = opencodeCfg(null, null, null);
-        // 5-arg (gate disabled) and 7-arg (gate enabled) production constructors both expose the profile.
-        OpenCodeLauncher disabled = new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of(cfg.profile(), cfg), cfg.profile(), _ -> null);
-        OpenCodeLauncher gated = new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of(cfg.profile(), cfg), cfg.profile(), _ -> null, 5000, 100);
-        assertEquals(java.util.Set.of("gemini"), disabled.profiles());
-        assertEquals("gemini", gated.defaultProfile());
-    }
-
-    @Test
-    void spawnGateThrowsPeerUnreachableWhenNeverInjectable(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        herdr.agentStatus("unknown"); // never injectable
-        long[] clock = {0};
-        OpenCodeLauncher svc = new OpenCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
-                Map.of("gemini", opencodeCfg(null, null, null)), "gemini", _ -> null,
-                1000, () -> clock[0], () -> clock[0] += 50, root, root);
-
-        PeerUnreachableException ex = assertThrows(PeerUnreachableException.class,
-                () -> svc.spawn(new SpawnRequest(null, null, null)));
-        assertTrue(clock[0] >= 1000, "the fake clock advanced past the timeout: " + clock[0]);
-        long closes = herdr.calls.stream()
-                .filter(c -> c.method().equals("pane.close"))
-                .filter(c -> "w9:pRoot_1".equals(((Map<?, ?>) c.params()).get("pane_id")))
-                .count();
-        assertEquals(1, closes, "the worker pane was reaped on timeout (no orphan)");
-        assertNotNull(ex.getMessage());
-    }
-
-    @Test
-    void spawnReturnsHandleWhenGateDisabled(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        PeerHandle handle = service(herdr, root, opencodeCfg(null, null, null))
-                .spawn(new SpawnRequest(null, null, null));
-        assertNotNull(handle, "spawn returns a handle when the gate is disabled");
-        assertFalse(herdr.called("agent.get"), "no polling when the gate is disabled");
-    }
-
-    @Test
-    void handleCarriesTheRealCharterReceiptNotTheInterfaceDefault(@TempDir Path root) {
-        // The base's WorkerHandle computes a real CharterReceipt (CB-571), but the opencode adapter
-        // wraps it in SessionAwareHandle for lazy session discovery. Before this fix that decorator
-        // did not override charterReceipt(), so it silently inherited PeerHandle's `null` default
-        // and the real receipt sitting on its delegate was lost.
-        FakeHerdr herdr = new FakeHerdr();
-        BridgedConfig.Fleet fleet = new BridgedConfig.Fleet(Map.of(), Map.of(), Map.of(), Map.of(),
-                Map.of("dev", "role rule"), null);
-        PeerHandle handle = service(herdr, root,
-                opencodeCfg("google/gemini-2.5-pro", "http://127.0.0.1:8765/mcp", null), () -> fleet)
-                .spawn(new SpawnRequest(null, null, null));
-
-        assertNotNull(handle.charterReceipt(),
-                "an opencode spawn's charterReceipt() must not silently be null");
-        String composed = "role rule\n\n" + HerdrPeerLauncher.REPLY_CHARTER;
-        assertEquals(CharterReceipt.digestOf(composed), handle.charterReceipt().charterSha256(),
-                "the receipt on the wrapped handle must match the exact composed charter bytes");
-    }
-
-    // --- CB-508: pinned OpenAI-compatible endpoint (e.g. a local vLLM) ---------------------------
-
-    /** A profile with a baseUrl but no model provider prefix cannot be resolved — fail loudly. */
-    private static BridgedConfig.Profile pinnedCfg(String model, String baseUrl, String mcpUrl) {
-        return new BridgedConfig.Profile("local", baseUrl, model, null, "BRIDGED_WORKER_TOKEN",
-                List.of("opencode"), "tab", "bridged-workers", "opencode: {model} #{n}", mcpUrl,
-                null, null, null, null, BridgedConfig.Profile.KIND_OPENCODE);
-    }
-
-    @Test
-    void baseUrlDeclaresACustomOpenAiCompatibleProvider(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, pinnedCfg("local-vllm/deepseek-v4-flash", "http://127.0.0.1:8000", null))
-                .spawn();
-
-        String cfgPath = startEnv(herdr).get("OPENCODE_CONFIG");
-        assertNotNull(cfgPath, "a pinned endpoint needs a config file even with no bridge MCP url");
-        JsonNode provider = new ObjectMapper().readTree(Path.of(cfgPath).toFile())
-                .path("provider").path("local-vllm");
-
-        assertFalse(provider.isMissingNode(), "the provider id comes from the model selector");
-        assertEquals("@ai-sdk/openai-compatible", provider.path("npm").asText());
-        assertEquals("http://127.0.0.1:8000/v1", provider.path("options").path("baseURL").asText(),
-                "a bare host:port gets /v1 appended — that is where these servers mount the API");
-        assertFalse(provider.path("options").path("apiKey").asText().isBlank(),
-                "the AI SDK requires a non-empty key even when the server ignores it");
-        assertFalse(provider.path("models").path("deepseek-v4-flash").isMissingNode(),
-                "the model half of the selector is declared under the provider");
-    }
-
-    @Test
-    void aBaseUrlThatAlreadyCarriesAPathIsUsedVerbatim(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, pinnedCfg("local-vllm/m", "http://127.0.0.1:8000/openai/v1", null)).spawn();
-
-        JsonNode json = new ObjectMapper()
-                .readTree(Path.of(startEnv(herdr).get("OPENCODE_CONFIG")).toFile());
-        assertEquals("http://127.0.0.1:8000/openai/v1",
-                json.path("provider").path("local-vllm").path("options").path("baseURL").asText(),
-                "an endpoint mounted on a custom path must not have /v1 bolted on");
-    }
-
-    @Test
-    void aPinnedEndpointRejectsAModelWithNoProviderPrefix(@TempDir Path root) {
-        FakeHerdr herdr = new FakeHerdr();
-        OpenCodeLauncher launcher =
-                service(herdr, root, pinnedCfg("deepseek-v4-flash", "http://127.0.0.1:8000", null));
-
-        // Silently falling back to the default gateway would point the worker at the wrong LLM
-        // while looking healthy — the one failure mode worth being loud about.
-        IllegalArgumentException e = assertThrows(IllegalArgumentException.class, launcher::spawn);
-        assertTrue(e.getMessage().contains("<provider>/<model>"), "the error says how to fix it");
-    }
-
-    @Test
-    void aPinnedEndpointAndTheBridgeMcpCoexistInOneConfig(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, pinnedCfg("local-vllm/deepseek-v4-flash",
-                "http://127.0.0.1:8000", "http://127.0.0.1:8766/mcp")).spawn();
-
-        JsonNode json = new ObjectMapper()
-                .readTree(Path.of(startEnv(herdr).get("OPENCODE_CONFIG")).toFile());
-        assertEquals("remote", json.path("mcp").path("bridge").path("type").asText(),
-                "pinning an endpoint must not drop the bridge MCP mount");
-        assertFalse(json.path("provider").path("local-vllm").isMissingNode(),
-                "and the provider block is still declared alongside it");
-        assertTrue(json.path("instructions").isArray() && !json.path("instructions").isEmpty(),
-                "the reply charter survives too");
-    }
-
-    @Test
-    void noBaseUrlDeclaresNoProviderSoTheDefaultGatewayIsUsed(@TempDir Path root) throws Exception {
-        FakeHerdr herdr = new FakeHerdr();
-        service(herdr, root, opencodeCfg("opencode/some-free-model", "http://127.0.0.1:8766/mcp", null))
-                .spawn();
-
-        JsonNode json = new ObjectMapper()
-                .readTree(Path.of(startEnv(herdr).get("OPENCODE_CONFIG")).toFile());
-        assertTrue(json.path("provider").isMissingNode(),
-                "without a baseUrl opencode resolves its own provider as before");
-    }
-}

bridged/src/test/java/dev/ltms/bridged/member/OpenCodeSessionDiscoveryTest.java

@@ -1,90 +0,0 @@
-package dev.ltms.bridged.member;
-
-import org.junit.jupiter.api.Test;
-import org.junit.jupiter.api.io.TempDir;
-
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.nio.file.attribute.FileTime;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-/**
- * {@link OpenCodeSessionDiscovery} matches an opencode session record by the worker's cwd (its
- * {@code directory}) against opencode's on-disk storage. These tests populate a TEMP storage root
- * themselves — never the operator's real {@code ~/.local/share/opencode}.
- */
-class OpenCodeSessionDiscoveryTest {
-
-    /**
-     * Write a session record {@code {"id":..., "directory":...}} under
-     * {@code <root>/session/<projectID>/<fileName>} and stamp it with a known last-modified time,
-     * so "most recently modified wins" is deterministic. Static so the launcher test can reuse it.
-     */
-    static void writeRecord(Path root, String projectId, String fileName, String id,
-                            String directory, long lastModifiedEpochMillis) throws Exception {
-        Path dir = root.resolve("session").resolve(projectId);
-        Files.createDirectories(dir);
-        Path file = dir.resolve(fileName);
-        Files.writeString(file, "{\"id\":\"" + id + "\",\"directory\":\"" + directory
-                + "\",\"projectID\":\"" + projectId + "\",\"version\":\"1.1.31\"}");
-        Files.setLastModifiedTime(file, FileTime.fromMillis(lastModifiedEpochMillis));
-    }
-
-    @Test
-    void findsTheRecordWhoseDirectoryEqualsTheCwd(@TempDir Path root) throws Exception {
-        writeRecord(root, "p1", "ses_a.json", "ses_aaa", "/w/a", 1000L);
-        writeRecord(root, "p2", "ses_b.json", "ses_bbb", "/w/b", 2000L);
-
-        assertEquals("ses_bbb", new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/b"),
-                "the record whose directory equals the cwd is the one found");
-        assertEquals("ses_aaa", new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/a"));
-    }
-
-    @Test
-    void aNonMatchingDirectoryYieldsNullRatherThanAMismatch(@TempDir Path root) throws Exception {
-        writeRecord(root, "p1", "ses_a.json", "ses_aaa", "/w/a", 1000L);
-
-        assertNull(new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/other"),
-                "no record for this cwd yet → null, not a wrong session");
-    }
-
-    @Test
-    void prefersTheMostRecentlyModifiedRecordWhenSeveralMatch(@TempDir Path root) throws Exception {
-        writeRecord(root, "p1", "old.json", "ses_old", "/w/a", 1000L);
-        writeRecord(root, "p2", "new.json", "ses_new", "/w/a", 5000L);
-
-        assertEquals("ses_new", new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/a"),
-                "the freshest record for the cwd wins");
-    }
-
-    @Test
-    void aMissingOrEmptyStorageRootYieldsNullWithoutThrowing(@TempDir Path root) throws Exception {
-        // Missing: no session dir at all under the root.
-        assertNull(new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/a"));
-
-        // Present but empty: a session dir with nothing in it produces no match, not a throw.
-        Path emptyRoot = root.resolve("empty");
-        Files.createDirectories(emptyRoot.resolve("session"));
-        assertNull(new OpenCodeSessionDiscovery(emptyRoot).sessionIdForDirectory("/w/a"));
-    }
-
-    @Test
-    void aBlankOrNullDirectoryYieldsNull(@TempDir Path root) {
-        OpenCodeSessionDiscovery discovery = new OpenCodeSessionDiscovery(root);
-        assertNull(discovery.sessionIdForDirectory(null));
-        assertNull(discovery.sessionIdForDirectory("   "));
-    }
-
-    @Test
-    void aMalformedRecordIsSkippedRatherThanFatal(@TempDir Path root) throws Exception {
-        // A record that fails to parse must not abort the scan of its siblings.
-        Path dir = root.resolve("session").resolve("p1");
-        Files.createDirectories(dir);
-        Files.writeString(dir.resolve("broken.json"), "{not valid json");
-        writeRecord(root, "p1", "good.json", "ses_good", "/w/a", 1000L);
-
-        assertEquals("ses_good", new OpenCodeSessionDiscovery(root).sessionIdForDirectory("/w/a"),
-                "an unreadable record is skipped; a later valid one still matches");
-    }
-}

bridged/src/test/java/dev/ltms/bridged/session/GitWorktreesTest.java

@@ -1,625 +0,0 @@
-package dev.ltms.bridged.session;
-
-import org.junit.jupiter.api.Test;
-import org.junit.jupiter.api.io.TempDir;
-
-import java.nio.charset.StandardCharsets;
-import java.nio.file.Files;
-import java.nio.file.Path;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Optional;
-import java.util.Set;
-import java.util.concurrent.TimeUnit;
-
-import static org.junit.jupiter.api.Assertions.*;
-
-/**
- * CB-525 acceptance test for tool-surface isolation. This is one of the few tests that drives real
- * {@code git} — the behaviour under test is precisely what {@link GitWorktrees} does to a checkout,
- * so a fake would assert nothing. Everything happens inside a {@link TempDir} throwaway repo.
- */
-class GitWorktreesTest {
-
-    /** A project MCP config with servers in it — what this repo actually commits. */
-    private static final String WITH_SERVERS = """
-            {
-              "mcpServers": {
-                "jetbrains": { "type": "sse", "url": "http://localhost:64342/sse" }
-              }
-            }
-            """;
-
-    /** An opencode config carrying a {@code {file:.secrets/...}} reference — CB-543's crash repro. */
-    private static final String OPENCODE_WITH_FILE_REF = """
-            {
-              "env": {
-                "CONTEXT7_TOKEN": "{file:.secrets/context7-token}"
-              }
-            }
-            """;
-
-    /** A non-empty autoenv file — the form that would prompt for authorization in a worktree. */
-    private static final String AUTOENV_WITH_DIRECTIVE = "export HELLO=world\n";
-
-    private static Path initRepo(Path dir) throws Exception {
-        Files.createDirectories(dir);
-        git(dir, "init", "-q", "-b", "main");
-        git(dir, "config", "user.email", "test@example.invalid");
-        git(dir, "config", "user.name", "Test");
-        Files.writeString(dir.resolve(".mcp.json"), WITH_SERVERS);
-        Files.writeString(dir.resolve("README.md"), "seed\n");
-        git(dir, "add", ".mcp.json", "README.md");
-        git(dir, "commit", "-q", "-m", "seed");
-        return dir;
-    }
-
-    private static void git(Path cwd, String... args) throws Exception {
-        List<String> cmd = new java.util.ArrayList<>(List.of("git"));
-        cmd.addAll(List.of(args));
-        Process p = new ProcessBuilder(cmd).directory(cwd.toFile()).redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git timed out: " + String.join(" ", cmd));
-        assertEquals(0, p.exitValue(), "git " + String.join(" ", args) + " failed:\n" + out);
-    }
-
-    /** Pending changes to {@code file} in {@code cwd}, empty when git considers it unmodified. */
-    private static String status(Path cwd, String file) throws Exception {
-        Process p = new ProcessBuilder("git", "status", "--porcelain", "--", file)
-                .directory(cwd.toFile()).redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git status timed out");
-        return out;
-    }
-
-    /** Every pending change in {@code cwd} — the whole-tree porcelain status, unlike {@link #status}. */
-    private static String fullStatus(Path cwd) throws Exception {
-        Process p = new ProcessBuilder("git", "status", "--porcelain")
-                .directory(cwd.toFile()).redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git status timed out");
-        return out;
-    }
-
-    private static String revParse(Path cwd, String ref) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "rev-parse", ref)
-                .redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes()).trim();
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git rev-parse timed out");
-        assertEquals(0, p.exitValue(), "git rev-parse " + ref + " failed:\n" + out);
-        return out;
-    }
-
-    /** The recursive file list of a commit's tree — used to check what a snapshot actually committed. */
-    private static String lsTree(Path cwd, String ref) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "ls-tree", "-r", "--name-only", ref)
-                .redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git ls-tree timed out");
-        assertEquals(0, p.exitValue(), "git ls-tree " + ref + " failed:\n" + out);
-        return out;
-    }
-
-    /** The set of paths in {@code git diff --name-only from..to} — used to check exactly what a
-     *  snapshot's tree changed relative to its parent, the same shape {@code git status --porcelain}
-     *  reports for the worktree it was taken from. */
-    private static Set<String> diffNameOnly(Path cwd, String from, String to) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "diff", "--name-only", from, to)
-                .redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git diff timed out");
-        assertEquals(0, p.exitValue(), "git diff " + from + ".." + to + " failed:\n" + out);
-        Set<String> paths = new HashSet<>();
-        for (String line : out.split("\\R")) {
-            if (!line.isBlank()) {
-                paths.add(line.trim());
-            }
-        }
-        return paths;
-    }
-
-    /** The set of paths a {@code git status --porcelain} listing names, stripping the two-char status
-     *  code prefix each line carries. */
-    private static Set<String> porcelainPaths(String porcelain) {
-        Set<String> paths = new HashSet<>();
-        for (String line : porcelain.split("\\R")) {
-            if (!line.isBlank()) {
-                paths.add(line.substring(3).trim());
-            }
-        }
-        return paths;
-    }
-
-    private static String forEachRef(Path cwd, String pattern) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "for-each-ref", pattern)
-                .redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes());
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git for-each-ref timed out");
-        assertEquals(0, p.exitValue(), "git for-each-ref " + pattern + " failed:\n" + out);
-        return out;
-    }
-
-    /** Write {@code content} as a blob into the object database; returns its sha. */
-    private static String blobOf(Path cwd, String content) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "hash-object", "-w", "--stdin")
-                .redirectErrorStream(true).start();
-        p.getOutputStream().write(content.getBytes(StandardCharsets.UTF_8));
-        p.getOutputStream().close();
-        String out = new String(p.getInputStream().readAllBytes()).trim();
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git hash-object timed out");
-        assertEquals(0, p.exitValue(), "git hash-object failed:\n" + out);
-        return out;
-    }
-
-    /** Build a single-file tree object from {@code blob}; returns the tree's sha. */
-    private static String treeOf(Path cwd, String path, String blob) throws Exception {
-        Process p = new ProcessBuilder("git", "-C", cwd.toString(), "mktree")
-                .redirectErrorStream(true).start();
-        p.getOutputStream().write(("100644 blob " + blob + "\t" + path + "\n").getBytes(StandardCharsets.UTF_8));
-        p.getOutputStream().close();
-        String out = new String(p.getInputStream().readAllBytes()).trim();
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git mktree timed out");
-        assertEquals(0, p.exitValue(), "git mktree failed:\n" + out);
-        return out;
-    }
-
-    /** {@code git commit-tree} rooted at {@code tree} with a chosen committer date; returns the sha. */
-    private static String commitTree(Path cwd, String tree, String parent, String committerDate,
-                                     String message) throws Exception {
-        ProcessBuilder pb = new ProcessBuilder("git", "-C", cwd.toString(), "commit-tree",
-                tree, "-p", parent, "-m", message);
-        pb.environment().put("GIT_COMMITTER_DATE", committerDate);
-        Process p = pb.redirectErrorStream(true).start();
-        String out = new String(p.getInputStream().readAllBytes()).trim();
-        assertTrue(p.waitFor(30, TimeUnit.SECONDS), "git commit-tree timed out");
-        assertEquals(0, p.exitValue(), "git commit-tree failed:\n" + out);
-        return out;
-    }
-
-    /** {@code git update-ref <ref> <sha>} — create the snapshot ref directly. */
-    private static void updateRef(Path cwd, String ref, String sha) throws Exception {
-        git(cwd, "update-ref", ref, sha);
-    }
-
-    /** True when {@code ref} exists in the repo (for-each-ref on a missing ref is empty, not an error). */
-    private static boolean refExists(Path cwd, String ref) throws Exception {
-        return !forEachRef(cwd, ref).trim().isEmpty();
-    }
-
-    /**
-     * The heart of CB-525: a provisioned worktree must not inherit the primary's MCP servers. Without
-     * the isolation step the checked-out {@code .mcp.json} carries them in, and a worker navigating
-     * through the primary's IDE servers edits the primary's tree while building its own.
-     */
-    @Test
-    void aProvisionedWorktreeInheritsNoMcpServers(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-525-a", "HEAD");
-
-        Path mcp = Path.of(wt).resolve(".mcp.json");
-        assertTrue(Files.exists(mcp), ".mcp.json must still exist — present and explicitly empty");
-        String body = Files.readString(mcp);
-        assertFalse(body.contains("jetbrains"), "worktree inherited the primary's MCP servers:\n" + body);
-        assertTrue(body.replaceAll("\\s+", "").contains("\"mcpServers\":{}"),
-                "expected an explicitly empty server map, got:\n" + body);
-    }
-
-    /** Neutralizing must not look like work in progress, or a worker would commit it into its PR. */
-    @Test
-    void theNeutralizedConfigIsNotAPendingLocalModification(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-525-b", "HEAD");
-
-        assertEquals("", status(Path.of(wt), ".mcp.json"),
-                "the neutralized .mcp.json shows as modified — --skip-worktree did not take");
-    }
-
-    /** Isolation is the worktree's business only; the primary's own checkout must be untouched. */
-    @Test
-    void thePrimaryCheckoutIsLeftAlone(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        new GitWorktrees(tmp.resolve("wts").toString()).add(repo.toString(), "cb-525-c", "HEAD");
-
-        assertEquals(WITH_SERVERS, Files.readString(repo.resolve(".mcp.json")),
-                "the primary's .mcp.json was rewritten — isolation reached out of the worktree");
-    }
-
-    /** A repo that commits no {@code .mcp.json} still gets one, so nothing can be inherited later. */
-    @Test
-    void aRepoWithoutAnMcpConfigStillGetsANeutralOne(@TempDir Path tmp) throws Exception {
-        Path repo = tmp.resolve("repo");
-        Files.createDirectories(repo);
-        git(repo, "init", "-q", "-b", "main");
-        git(repo, "config", "user.email", "test@example.invalid");
-        git(repo, "config", "user.name", "Test");
-        Files.writeString(repo.resolve("README.md"), "seed\n");
-        git(repo, "add", "README.md");
-        git(repo, "commit", "-q", "-m", "seed");
-
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-525-d", "HEAD");
-
-        // Untracked is the normal case here, so the --skip-worktree branch must be skipped rather
-        // than run and fail: `update-index --skip-worktree` on an unknown path exits non-zero.
-        String body = Files.readString(Path.of(wt).resolve(".mcp.json"));
-        assertTrue(body.replaceAll("\\s+", "").contains("\"mcpServers\":{}"), body);
-    }
-
-    /**
-     * CB-543's repro: a tracked {@code opencode.json} carries a {@code {file:.secrets/...}} reference
-     * to a gitignored secret that never reaches a worktree, and opencode refuses to start on it. The
-     * worktree's copy must be neutralized and hidden like {@code .mcp.json}.
-     */
-    @Test
-    void aTrackedOpencodeConfigIsNeutralizedAndHidden(@TempDir Path tmp) throws Exception {
-        Path repo = tmp.resolve("repo");
-        Files.createDirectories(repo);
-        git(repo, "init", "-q", "-b", "main");
-        git(repo, "config", "user.email", "test@example.invalid");
-        git(repo, "config", "user.name", "Test");
-        Files.writeString(repo.resolve(".mcp.json"), WITH_SERVERS);
-        Files.writeString(repo.resolve("opencode.json"), OPENCODE_WITH_FILE_REF);
-        Files.writeString(repo.resolve("README.md"), "seed\n");
-        git(repo, "add", ".mcp.json", "opencode.json", "README.md");
-        git(repo, "commit", "-q", "-m", "seed");
-
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-543-a", "HEAD");
-
-        String body = Files.readString(Path.of(wt).resolve("opencode.json"));
-        assertFalse(body.contains(".secrets"),
-                "worktree kept a dangling {file:...} secret reference:\n" + body);
-        assertEquals("{}", body.replaceAll("\\s+", ""),
-                "expected an empty JSON object stub, got:\n" + body);
-        assertEquals("", status(Path.of(wt), "opencode.json"),
-                "the neutralized opencode.json shows as modified — --skip-worktree did not take");
-    }
-
-    /** A config the repo does not carry must be skipped — no stub invented, provisioning still succeeds. */
-    @Test
-    void anAbsentConfigIsSkippedWithoutError(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));   // only .mcp.json + README are committed
-
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-543-b", "HEAD");
-
-        assertFalse(Files.exists(Path.of(wt).resolve("opencode.json")),
-                "a stub was invented for a config the repo does not carry");
-        assertFalse(Files.exists(Path.of(wt).resolve(".autoenv")),
-                "a stub was invented for a config the repo does not carry");
-        // .mcp.json's long-standing create-always behaviour must be unchanged.
-        assertTrue(Files.exists(Path.of(wt).resolve(".mcp.json")), ".mcp.json stub was dropped");
-    }
-
-    /**
-     * CB-576. {@code hasUncommitted} must treat a freshly-provisioned worktree as clean, but a
-     * worktree holding a brand-new, never-added file as dirty. The untracked-file-only shape is
-     * exactly the work lost in the incident — a worker's draft that compiled but was never
-     * committed because it stopped to ask its lead a question.
-     */
-    @Test
-    void anUntrackedOnlyWorktreeCountsAsDirty(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String wt = gitWorktrees.add(repo.toString(), "cb-576-u", "HEAD");
-
-        assertFalse(gitWorktrees.hasUncommitted(wt),
-                "a freshly provisioned worktree must read as clean");
-
-        Files.writeString(Path.of(wt).resolve("brand-new.txt"), "draft that was never added\n");
-
-        assertTrue(gitWorktrees.hasUncommitted(wt),
-                "an untracked-only file must count as dirty");
-
-        Files.writeString(Path.of(wt).resolve("README.md"), "edited tracked file\n");
-        assertTrue(gitWorktrees.hasUncommitted(wt),
-                "a tracked modification must also count as dirty");
-    }
-
-    /**
-     * CB-576 review. {@code hasUncommitted} must tolerate a missing worktree exactly like
-     * {@code remove}: an already-gone directory holds no work to lose, and throwing here would
-     * break teardown — SessionManager.release() calls it before stopping the pane, so an
-     * exception would orphan a live pane and skip the release notification (CB-516).
-     */
-    @Test
-    void hasUncommittedOnAMissingWorktreeReturnsFalseWithoutThrowing(@TempDir Path tmp) {
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String gone = tmp.resolve("wts").resolve("does-not-exist").toString();
-
-        assertFalse(gitWorktrees.hasUncommitted(gone),
-                "a missing worktree is reported clean, not an error");
-    }
-
-    /** All three protected configs are covered: each one present in a worktree is neutralized and hidden. */
-    @Test
-    void allThreeConfigsAreNeutralizedWhenPresent(@TempDir Path tmp) throws Exception {
-        Path repo = tmp.resolve("repo");
-        Files.createDirectories(repo);
-        git(repo, "init", "-q", "-b", "main");
-        git(repo, "config", "user.email", "test@example.invalid");
-        git(repo, "config", "user.name", "Test");
-        Files.writeString(repo.resolve(".mcp.json"), WITH_SERVERS);
-        Files.writeString(repo.resolve("opencode.json"), OPENCODE_WITH_FILE_REF);
-        Files.writeString(repo.resolve(".autoenv"), AUTOENV_WITH_DIRECTIVE);
-        Files.writeString(repo.resolve("README.md"), "seed\n");
-        git(repo, "add", ".mcp.json", "opencode.json", ".autoenv", "README.md");
-        git(repo, "commit", "-q", "-m", "seed");
-
-        String wt = new GitWorktrees(tmp.resolve("wts").toString())
-                .add(repo.toString(), "cb-543-c", "HEAD");
-
-        assertTrue(Files.readString(Path.of(wt).resolve(".mcp.json"))
-                        .replaceAll("\\s+", "").contains("\"mcpServers\":{}"),
-                ".mcp.json was not neutralized");
-        assertEquals("{}", Files.readString(Path.of(wt).resolve("opencode.json")).replaceAll("\\s+", ""),
-                "opencode.json was not neutralized");
-        assertEquals("", Files.readString(Path.of(wt).resolve(".autoenv")),
-                ".autoenv was not neutralized");
-
-        assertEquals("", status(Path.of(wt), ".mcp.json"), ".mcp.json still shows as modified");
-        assertEquals("", status(Path.of(wt), "opencode.json"), "opencode.json still shows as modified");
-        assertEquals("", status(Path.of(wt), ".autoenv"), ".autoenv still shows as modified");
-    }
-
-    /**
-     * CB-578 stage C, acceptance criterion 1. A dirty worktree — a tracked edit plus a brand-new
-     * untracked file, exactly the shape lost in CB-576 — must land in {@code refs/wip/<branch>}'s
-     * tree, and that ref must live outside {@code refs/heads} so it never shows up in
-     * {@code git branch} or gets swept by a branch cleanup.
-     */
-    @Test
-    void dirtySnapshotCreatesARefWhoseTreeContainsUntrackedAndTrackedChanges(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-578-c-a";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-        Files.writeString(Path.of(wt).resolve("untracked.txt"), "draft that was never added\n");
-        Files.writeString(Path.of(wt).resolve("README.md"), "edited tracked file\n");
-
-        Optional<String> ref = gitWorktrees.snapshot(wt, branch, "test snapshot");
-
-        assertTrue(ref.isPresent(), "a dirty worktree snapshot returns a commit sha");
-        String tree = lsTree(repo, "refs/wip/" + branch);
-        assertTrue(tree.contains("untracked.txt"), "snapshot tree must include the untracked file:\n" + tree);
-        assertTrue(tree.contains("README.md"), "snapshot tree must include the tracked edit:\n" + tree);
-        String heads = forEachRef(repo, "refs/heads");
-        assertFalse(heads.contains("refs/wip/"), "the snapshot ref must not live under refs/heads:\n" + heads);
-    }
-
-    /**
-     * CB-578 stage C, acceptance criterion 2. Building the commit through a temporary
-     * {@code GIT_INDEX_FILE} must leave the worker's own index, working tree, and HEAD exactly as
-     * they were — the worker may still be mid-write, and staging into the real index would corrupt
-     * that.
-     */
-    @Test
-    void snapshotDoesNotTouchTheWorkersOwnIndexWorkingTreeOrHead(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-578-c-b";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-        Files.writeString(Path.of(wt).resolve("untracked.txt"), "draft that was never added\n");
-        Files.writeString(Path.of(wt).resolve("README.md"), "edited tracked file\n");
-        String headBefore = revParse(Path.of(wt), "HEAD");
-        String statusBefore = fullStatus(Path.of(wt));
-
-        gitWorktrees.snapshot(wt, branch, "test snapshot");
-
-        assertEquals(headBefore, revParse(Path.of(wt), "HEAD"), "snapshot must not move HEAD");
-        assertEquals(statusBefore, fullStatus(Path.of(wt)),
-                "snapshot must not change the worker's own index or working tree status");
-    }
-
-    /**
-     * CB-578 stage C, acceptance criterion 3. {@code add -A} (never {@code -f}) respects
-     * {@code .gitignore}, and that is the only thing keeping a gitignored file (secrets, local
-     * config) out of a snapshot commit — an untracked file that is NOT ignored must still be
-     * included, so this isn't just "untracked files are dropped".
-     */
-    @Test
-    void gitignoredFileIsExcludedButOtherUntrackedFilesAreNot(@TempDir Path tmp) throws Exception {
-        Path repo = tmp.resolve("repo");
-        Files.createDirectories(repo);
-        git(repo, "init", "-q", "-b", "main");
-        git(repo, "config", "user.email", "test@example.invalid");
-        git(repo, "config", "user.name", "Test");
-        Files.writeString(repo.resolve(".gitignore"), ".env\n");
-        Files.writeString(repo.resolve("README.md"), "seed\n");
-        git(repo, "add", ".gitignore", "README.md");
-        git(repo, "commit", "-q", "-m", "seed");
-
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-578-c-c";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-        Files.writeString(Path.of(wt).resolve(".env"), "SECRET=shh\n");
-        Files.writeString(Path.of(wt).resolve("untracked.txt"), "draft that was never added\n");
-
-        Optional<String> ref = gitWorktrees.snapshot(wt, branch, "test snapshot");
-
-        assertTrue(ref.isPresent());
-        String tree = lsTree(repo, "refs/wip/" + branch);
-        assertFalse(tree.contains(".env"), "a gitignored file must never enter the snapshot:\n" + tree);
-        assertTrue(tree.contains("untracked.txt"),
-                "a non-ignored untracked file must still be included:\n" + tree);
-    }
-
-    /** CB-578 stage C. A clean worktree still produces a valid, if tree-identical, commit — the caller
-     *  (SessionManager) is the one that decides not to call this on a clean worktree. */
-    @Test
-    void snapshotOfAMissingWorktreeReturnsEmptyWithoutThrowing(@TempDir Path tmp) {
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String gone = tmp.resolve("wts").resolve("does-not-exist").toString();
-
-        assertTrue(gitWorktrees.snapshot(gone, "some-branch", "msg").isEmpty(),
-                "a missing worktree has nothing to snapshot, and must not throw");
-    }
-
-    /**
-     * CB-587, acceptance criteria 1 and 2. A file with a REAL {@code --skip-worktree} bit set in the
-     * worktree's own index must never enter the snapshot, even though its on-disk content has locally
-     * diverged from what is committed — that is exactly the divergence {@code --skip-worktree} exists
-     * to hide from {@code git status}, and a snapshot built from a fresh empty temp index (the bug)
-     * stages that local content anyway because the fresh index carries none of the real index's flags.
-     * The snapshot's diff against its parent must list exactly what {@code git status --porcelain}
-     * reports for the worktree — no more, no less.
-     */
-    @Test
-    void dirtySnapshotHonoursARealSkipWorktreeBit(@TempDir Path tmp) throws Exception {
-        Path repo = tmp.resolve("repo");
-        Files.createDirectories(repo);
-        git(repo, "init", "-q", "-b", "main");
-        git(repo, "config", "user.email", "test@example.invalid");
-        git(repo, "config", "user.name", "Test");
-        Files.writeString(repo.resolve("protected.cfg"), "committed-value\n");
-        Files.writeString(repo.resolve("README.md"), "seed\n");
-        git(repo, "add", "protected.cfg", "README.md");
-        git(repo, "commit", "-q", "-m", "seed");
-
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-587-a";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-
-        git(Path.of(wt), "update-index", "--skip-worktree", "protected.cfg");
-        Files.writeString(Path.of(wt).resolve("protected.cfg"), "locally-diverged-never-commit\n");
-        Files.writeString(Path.of(wt).resolve("untracked.txt"), "draft that was never added\n");
-        Files.writeString(Path.of(wt).resolve("README.md"), "edited tracked file\n");
-
-        String porcelain = fullStatus(Path.of(wt));
-        assertFalse(porcelain.contains("protected.cfg"),
-                "test setup invalid — protected.cfg must not show in git status once skip-worktree is set:\n"
-                        + porcelain);
-
-        Optional<String> ref = gitWorktrees.snapshot(wt, branch, "test snapshot");
-
-        assertTrue(ref.isPresent(), "a dirty worktree snapshot returns a commit sha");
-        Set<String> diffPaths = diffNameOnly(repo, "HEAD", "refs/wip/" + branch);
-        assertFalse(diffPaths.contains("protected.cfg"),
-                "a --skip-worktree file's local drift leaked into the snapshot:\n" + diffPaths);
-        assertEquals(porcelainPaths(porcelain), diffPaths,
-                "snapshot diff must list exactly what git status --porcelain reports, no more, no less");
-    }
-
-    /**
-     * CB-587, acceptance criterion 6. If the worktree's real index cannot be resolved/read, snapshot
-     * must fail loudly (throw) rather than silently falling back to an empty temp index and producing
-     * a wrong snapshot. SessionManager's caller already catches and WARNs on any exception here — this
-     * only needs to confirm the failure is not swallowed inside snapshot() itself.
-     */
-    @Test
-    void snapshotThrowsWhenTheWorktreesRealIndexCannotBeResolved(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-587-b";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-
-        // Break git's ability to resolve the worktree's real index by removing the linked worktree's
-        // `.git` file (which normally points at the main repo's worktrees/<name>/ directory).
-        Files.delete(Path.of(wt).resolve(".git"));
-
-        assertThrows(WorktreeException.class, () -> gitWorktrees.snapshot(wt, branch, "test snapshot"),
-                "an unresolvable real index must fail loudly, not silently snapshot from an empty index");
-    }
-
-    /**
-     * CB-586, criterion 2. A snapshot whose content is NOT reachable from {@code main} is the last
-     * copy of a worker's work, and must never be deleted automatically — even when it is old and
-     * even when the caller passes a zero age floor. Uses the real snapshot path on a dirty worktree,
-     * so the unreachable tree is exactly the shape CB-576/CB-578 stage C exist to protect.
-     */
-    @Test
-    void anUnreachableSnapshotIsNeverPruned(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-        String branch = "cb-586-unreachable";
-        String wt = gitWorktrees.add(repo.toString(), branch, "HEAD");
-        Files.writeString(Path.of(wt).resolve("worker-draft.txt"), "work that exists nowhere else\n");
-
-        Optional<String> ref = gitWorktrees.snapshot(wt, branch, "snapshot with unreachable content");
-        assertTrue(ref.isPresent());
-
-        // Age floor 0 makes age a non-issue: only reachability can save it — and it must.
-        assertEquals(0, gitWorktrees.pruneWipRefs(repo.toString(), 0),
-                "the unreachable snapshot is the last copy and must not be pruned");
-        assertTrue(refExists(repo, "refs/wip/" + branch),
-                "an unreachable snapshot must survive the sweep");
-    }
-
-    /**
-     * CB-586, criterion 1 (the reachable half). A snapshot whose tree content IS already reachable
-     * from {@code main} and which is older than the age floor is pure duplication — the work is
-     * recovered — so it must be pruned.
-     */
-    @Test
-    void aReachableSnapshotOlderThanTheFloorIsPruned(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-
-        // A snapshot whose tree is exactly main's current tree: fully reachable from main.
-        String mainTree = revParse(repo, "main^{tree}");
-        String old = commitTree(repo, mainTree, revParse(repo, "HEAD"), "2020-01-01T00:00:00", "snapshot");
-        updateRef(repo, "refs/wip/recovered", old);
-
-        assertEquals(1, gitWorktrees.pruneWipRefs(repo.toString(), TimeUnit.HOURS.toMillis(24)),
-                "an old, main-reachable snapshot must be pruned");
-        assertFalse(refExists(repo, "refs/wip/recovered"),
-                "the reachable snapshot's ref must be gone after the sweep");
-    }
-
-    /**
-     * CB-586, the age floor. A snapshot whose content IS reachable from {@code main} but which is
-     * younger than the age floor must not be swept — a lead may still be looking at it.
-     */
-    @Test
-    void aReachableButRecentSnapshotIsNotPruned(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-
-        // Reachable from main, but committed "now" — a fresh snapshot. The 24h floor must protect it.
-        String mainTree = revParse(repo, "main^{tree}");
-        String fresh = commitTree(repo, mainTree, revParse(repo, "HEAD"),
-                "2038-01-01T00:00:00", "snapshot just taken");
-        updateRef(repo, "refs/wip/fresh", fresh);
-
-        assertEquals(0, gitWorktrees.pruneWipRefs(repo.toString(), TimeUnit.HOURS.toMillis(24)),
-                "a recent snapshot must be kept even when reachable");
-        assertTrue(refExists(repo, "refs/wip/fresh"),
-                "the recent reachable snapshot must survive the sweep");
-    }
-
-    /**
-     * CB-586, criterion 5. A fleet that has never snapshotted anything has no {@code refs/wip/*},
-     * so a sweep is a no-op and the census reports none — identical to before CB-586 existed.
-     */
-    @Test
-    void aFleetWithNoSnapshotsPrunesNothingAndReportsNothing(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-
-        assertEquals(0, gitWorktrees.pruneWipRefs(repo.toString(), 0),
-                "no snapshot refs means nothing to prune");
-        Worktrees.WipRefStats stats = gitWorktrees.wipRefs(repo.toString());
-        assertEquals(0, stats.count(), "a never-snapshotted fleet has zero refs/wip refs");
-        assertEquals(0L, stats.costBytes(), "a never-snapshotted fleet costs zero bytes");
-    }
-
-    /** CB-586, criterion 4: the census reports how many refs exist and roughly what they cost. */
-    @Test
-    void wipRefsReportsCountAndCost(@TempDir Path tmp) throws Exception {
-        Path repo = initRepo(tmp.resolve("repo"));
-        GitWorktrees gitWorktrees = new GitWorktrees(tmp.resolve("wts").toString());
-
-        String blob = blobOf(repo, "a recoverable snapshot's worth of content");
-        String tree = treeOf(repo, "snapshot.txt", blob);
-        updateRef(repo, "refs/wip/one", commitTree(repo, tree, revParse(repo, "HEAD"),
-                "2020-01-01T00:00:00", "snapshot"));
-        updateRef(repo, "refs/wip/two", commitTree(repo, tree, revParse(repo, "HEAD"),
-                "2020-01-02T00:00:00", "snapshot"));
-
-        Worktrees.WipRefStats stats = gitWorktrees.wipRefs(repo.toString());
-        assertEquals(2, stats.count(), "two snapshot refs are reported");
-        assertTrue(stats.costBytes() > 0, "the cost of the snapshots is a positive byte count");
-    }
-}

deploy/dev.ltms.fleetd.plist

@@ -1,36 +1,36 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
 <!--
-  CB-504 / CB-594 — launchd agent for bridged (macOS).
+  CB-504 / CB-594 — launchd agent for fleetd (macOS).
 
   This is the real supervision target today: the dogfooded daemon runs on macOS, where there is
-  no systemd. A systemd unit ships alongside (deploy/bridged.service) for the Linux gateways
+  no systemd. A systemd unit ships alongside (deploy/fleetd.service) for the Linux gateways
   CB-308 introduces.
 
   Install:
-    cp deploy/dev.ltms.bridged.plist ~/Library/LaunchAgents/
-    launchctl load -w ~/Library/LaunchAgents/dev.ltms.bridged.plist
-    launchctl list | grep bridged
+    cp deploy/dev.ltms.fleetd.plist ~/Library/LaunchAgents/
+    launchctl load -w ~/Library/LaunchAgents/dev.ltms.fleetd.plist
+    launchctl list | grep fleetd
 
   The paths below are already filled in for this host (resolved 2026-08-16 from
   `/usr/libexec/java_home`... except that reported the system Applet-plugin JVM, not the jenv-
-  managed JDK 25 actually used to build/run bridged, so JAVA_HOME here is the real one:
+  managed JDK 25 actually used to build/run fleetd, so JAVA_HOME here is the real one:
   `JENV_VERSION=25.0.3 java -XshowSettings:properties -version 2>&1 | grep java.home`; `which mvn`;
   `echo $HOME`). If this file is copied to a different host, re-resolve all three paths and check
-  no placeholder path is left behind; scripts/redeploy-bridged.sh's check mode does not (and
+  no placeholder path is left behind; scripts/redeploy-fleetd.sh's check mode does not (and
   cannot) check this file for you.
 
   CB-594 — launchd cannot run a login shell (see the PATH comment on EnvironmentVariables below,
-  and scripts/bridged-launchd-wrapper.sh for the fix): ProgramArguments below execs THAT wrapper,
+  and scripts/fleetd-launchd-wrapper.sh for the fix): ProgramArguments below execs THAT wrapper,
   not java directly, so WORKER_GITEA_TOKEN and AI_GATEWAY_TOKEN still get sourced from
   ${SHARED_ENV}/tools/secrets.sh even though launchd itself never sources anything.
 
   Note on ordering: launchd has no "start after herdr" primitive for user agents, and neither
-  does systemd in a way that survives a socket appearing late. bridged retries the herdr socket
+  does systemd in a way that survives a socket appearing late. fleetd retries the herdr socket
   on startup instead, so an agent that comes up before herdr converges rather than dying — that
   retry is the actual fix; KeepAlive below is the backstop.
 
-  CB-594 — KeepAlive vs. scripts/redeploy-bridged.sh: a bare SIGTERM makes this JVM exit 143 even
+  CB-594 — KeepAlive vs. scripts/redeploy-fleetd.sh: a bare SIGTERM makes this JVM exit 143 even
   with its shutdown hook running to completion (measured, see the CB-594 report), which
   SuccessfulExit:false below reads as a crash and races to restart the OLD jar. The redeploy
   script now detects a loaded agent and uses `launchctl unload`/`load` instead of a raw kill, so
@@ -40,20 +40,20 @@
 <plist version="1.0">
 <dict>
     <key>Label</key>
-    <string>dev.ltms.bridged</string>
+    <string>dev.ltms.fleetd</string>
 
     <key>ProgramArguments</key>
     <array>
-        <string>/Users/dai.ha/LTMS/claude-bridge/scripts/bridged-launchd-wrapper.sh</string>
+        <string>/Users/dai.ha/LTMS/claude-bridge/scripts/fleetd-launchd-wrapper.sh</string>
         <string>/Users/dai.ha/Softwares/jdks/jdk-25.0.3.jdk/Contents/Home/bin/java</string>
         <string>-jar</string>
-        <string>/Users/dai.ha/LTMS/claude-bridge/bridged/target/bridged.jar</string>
-        <string>bridged.yaml</string>
+        <string>/Users/dai.ha/LTMS/claude-bridge/fleetd/target/fleetd.jar</string>
+        <string>fleetd.yaml</string>
     </array>
 
     <!-- Config path in ProgramArguments is relative, so the working directory must be the module. -->
     <key>WorkingDirectory</key>
-    <string>/Users/dai.ha/LTMS/claude-bridge/bridged</string>
+    <string>/Users/dai.ha/LTMS/claude-bridge/fleetd</string>
 
     <key>EnvironmentVariables</key>
     <dict>
@@ -62,7 +62,7 @@
         <key>HERDR_SOCKET_PATH</key>
         <string>/Users/dai.ha/.config/herdr/herdr.sock</string>
         <!--
-          PATH matters more than it looks (CB-511): bridged propagates its own PATH to every worker
+          PATH matters more than it looks (CB-511): fleetd propagates its own PATH to every worker
           it spawns, so this line decides whether the fleet can run a build at all. launchd does NOT
           source .zprofile/.zshrc, so without this the daemon (and therefore every worker) gets a
           bare /usr/bin:/bin and no JDK or Maven. Keep the toolchain entries first.
@@ -71,10 +71,10 @@
         <string>/Users/dai.ha/Softwares/jdks/jdk-25.0.3.jdk/Contents/Home/bin:/Users/dai.ha/Softwares/apache-maven/bin:/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin</string>
         <!--
           Worker/API tokens are NOT set here: this file is committed. CB-594 —
-          scripts/bridged-launchd-wrapper.sh (named in ProgramArguments above) is what supplies
+          scripts/fleetd-launchd-wrapper.sh (named in ProgramArguments above) is what supplies
           them, by execing a login shell that sources ${SHARED_ENV}/tools/secrets.sh before the
-          daemon itself starts. bridged also reads the API token from the env var named by
-          auth.tokenEnv (default BRIDGED_API_TOKEN) and only in auth.mode: token — the wrapper
+          daemon itself starts. fleetd also reads the API token from the env var named by
+          auth.tokenEnv (default FLEETD_API_TOKEN) and only in auth.mode: token — the wrapper
           covers that one too, since it is the same login shell.
         -->
     </dict>
@@ -84,21 +84,21 @@
 
     <!--
       CB-600 — read this before assuming ThrottleInterval bounds anything. It paces restarts to at
-      most one per 10s; it does NOT cap how many times launchd retries. If bridged fails fast on
-      every start — a bad bridged.yaml, for example auth.mode: token with the token env var unset,
+      most one per 10s; it does NOT cap how many times launchd retries. If fleetd fails fast on
+      every start — a bad fleetd.yaml, for example auth.mode: token with the token env var unset,
       which throws in main() before the daemon ever binds a port — launchd restarts it forever,
       once every 10s, until a human intervenes. LaunchAgents have no "give up after N attempts"
       primitive, so this is not something a config change here can fix.
 
-      That loop stops only two ways: (1) `launchctl unload -w ~/Library/LaunchAgents/dev.ltms.bridged.plist`,
+      That loop stops only two ways: (1) `launchctl unload -w ~/Library/LaunchAgents/dev.ltms.fleetd.plist`,
       or (2) the underlying cause gets fixed, so the process starts successfully and stays up (no
-      more exits to restart). scripts/redeploy-bridged.sh does not add a third way — it does not
-      make bridged self-disable on a config error, on purpose: a fail-fast exit path that
+      more exits to restart). scripts/redeploy-fleetd.sh does not add a third way — it does not
+      make fleetd self-disable on a config error, on purpose: a fail-fast exit path that
       sometimes decides "this is unrecoverable, stop trying" is one more thing that can misfire,
       and a wrongly self-disabled daemon needs the exact same manual `launchctl load -w` recovery
       this comment already names — so it buys nothing an operator watching for the crash loop
       doesn't already have, at the cost of a new way to be silently down. Watch for it with
-      `launchctl list dev.ltms.bridged` (a high restart count) or by tailing bridged.out for the
+      `launchctl list dev.ltms.fleetd` (a high restart count) or by tailing fleetd.out for the
       same startup error repeating every ~10s.
     -->
     <key>KeepAlive</key>
@@ -110,16 +110,16 @@
     <integer>10</integer>
 
     <!--
-      CB-594 — same file scripts/redeploy-bridged.sh already tails ($BRIDGED/bridged.out), and both
+      CB-594 — same file scripts/redeploy-fleetd.sh already tails ($BRIDGED/fleetd.out), and both
       streams point at it, not two separate log files: the script's fresh-line / ERROR-count checks
       after a restart read this one path regardless of whether launchd or the script started the
       process, and a stdout/stderr split would make half of what happens during a launchd-driven
       restart invisible to it.
     -->
     <key>StandardOutPath</key>
-    <string>/Users/dai.ha/LTMS/claude-bridge/bridged/bridged.out</string>
+    <string>/Users/dai.ha/LTMS/claude-bridge/fleetd/fleetd.out</string>
     <key>StandardErrorPath</key>
-    <string>/Users/dai.ha/LTMS/claude-bridge/bridged/bridged.out</string>
+    <string>/Users/dai.ha/LTMS/claude-bridge/fleetd/fleetd.out</string>
 
     <key>ProcessType</key>
     <string>Background</string>

deploy/fleetd.service

@@ -1,45 +1,45 @@
-# CB-504 — systemd unit for bridged (Linux).
+# CB-504 — systemd unit for fleetd (Linux).
 #
-# The macOS launchd agent (deploy/dev.ltms.bridged.plist) is the supervision target for the
+# The macOS launchd agent (deploy/dev.ltms.fleetd.plist) is the supervision target for the
 # current single-host deployment. This unit exists for the per-host gateways CB-308 introduces,
 # which will run on Linux.
 #
-# Install (user service — bridged drives the user's herdr, not a system daemon):
+# Install (user service — fleetd drives the user's herdr, not a system daemon):
 #   mkdir -p ~/.config/systemd/user
-#   cp deploy/bridged.service ~/.config/systemd/user/
+#   cp deploy/fleetd.service ~/.config/systemd/user/
 #   # edit ExecStart / WorkingDirectory / Environment below, then:
 #   systemctl --user daemon-reload
-#   systemctl --user enable --now bridged
-#   journalctl --user -u bridged -f
+#   systemctl --user enable --now fleetd
+#   journalctl --user -u fleetd -f
 
 [Unit]
-Description=bridged — claude-bridge message server
-Documentation=https://git.ltms.dev/lms/claude-bridge/wiki
+Description=fleetd — claude-bridge message server
+Documentation=https://git.ltms.dev/fleet/fleetd/wiki
 # Ordering only: herdr is a user process and its socket may appear after us. This is advisory —
-# bridged retries the herdr socket rather than exiting, which is what actually makes a late
-# socket survivable. Do NOT add Requires=: a herdr restart must not take bridged down with it.
+# fleetd retries the herdr socket rather than exiting, which is what actually makes a late
+# socket survivable. Do NOT add Requires=: a herdr restart must not take fleetd down with it.
 After=herdr.service
 Wants=herdr.service
 
 [Service]
 Type=simple
-WorkingDirectory=%h/src/claude-bridge/bridged
-ExecStart=/usr/lib/jvm/temurin-25-jdk/bin/java -jar target/bridged.jar bridged.yaml
+WorkingDirectory=%h/src/claude-bridge/fleetd
+ExecStart=/usr/lib/jvm/temurin-25-jdk/bin/java -jar target/fleetd.jar fleetd.yaml
 
 Environment=HERDR_SOCKET_PATH=%h/.config/herdr/herdr.sock
-# PATH matters more than it looks (CB-511): bridged propagates its own PATH to every worker it
+# PATH matters more than it looks (CB-511): fleetd propagates its own PATH to every worker it
 # spawns, so this line decides whether the fleet can run a build at all. systemd does not source a
 # login shell, so without it the daemon — and every worker — gets a bare default with no JDK/Maven.
 Environment=PATH=/usr/lib/jvm/temurin-25-jdk/bin:/usr/share/maven/bin:/usr/local/bin:/usr/bin:/bin
 # Secrets are NOT set here — this file is committed. Put the API/worker tokens in a private
 # drop-in that systemd reads with restrictive permissions:
-#   systemctl --user edit bridged     →  [Service] / Environment=BRIDGED_API_TOKEN=...
+#   systemctl --user edit fleetd     →  [Service] / Environment=FLEETD_API_TOKEN=...
 # or point EnvironmentFile at a 0600 file:
-# EnvironmentFile=%h/.config/bridged/env
+# EnvironmentFile=%h/.config/fleetd/env
 
 Restart=on-failure
 RestartSec=10s
-# A bad config (e.g. a non-loopback bind without token auth) makes bridged fail fast by design.
+# A bad config (e.g. a non-loopback bind without token auth) makes fleetd fail fast by design.
 # Give up rather than restart-loop on a permanent error.
 StartLimitBurst=5
 StartLimitIntervalSec=120
@@ -55,7 +55,7 @@ RestrictSUIDSGID=true
 
 StandardOutput=journal
 StandardError=journal
-SyslogIdentifier=bridged
+SyslogIdentifier=fleetd
 
 [Install]
 WantedBy=default.target

deploy/lavinmq/compose.yaml

@@ -1,8 +1,8 @@
-# LavinMQ — the AMQP broker behind bridged's durable ReplyInbox (CB-307 Stage 2).
+# LavinMQ — the AMQP broker behind fleetd's durable ReplyInbox (CB-307 Stage 2).
 #
 # Why this file exists: the broker was previously run ad hoc and simply vanished from the host,
-# which takes bridged down with it — AmqpReplyInbox.open throws on an unreachable broker and
-# Bridged.java:187 does not guard it, so a missing broker is a hard startup failure, not a
+# which takes fleetd down with it — AmqpReplyInbox.open throws on an unreachable broker and
+# Fleetd.java:187 does not guard it, so a missing broker is a hard startup failure, not a
 # degraded mode. This pins the version, keeps the data, and brings itself back after a reboot.
 #
 # Usage:
@@ -14,27 +14,27 @@
 #
 # Management UI: http://127.0.0.1:15672  (guest / guest)
 #
-# This is bridged's OWN broker. Do not point bridged at any other AMQP server on this host —
+# This is fleetd's OWN broker. Do not point fleetd at any other AMQP server on this host —
 # notably not the `local-rabbitmq` container, which belongs to a different project and would end
 # up carrying this project's queues.
 
-name: bridged-broker
+name: fleetd-broker
 
 services:
   lavinmq:
     # Pinned deliberately: :latest silently moves the broker under a running daemon.
     image: cloudamqp/lavinmq:2.9.1
-    container_name: bridged-lavinmq
+    container_name: fleetd-lavinmq
 
     # The failure this deployment exists to prevent — survive reboots and Docker restarts, but
     # stay down if it was stopped on purpose.
     restart: unless-stopped
 
     # Loopback-bound on purpose. LavinMQ ships a default guest/guest account, which is only
-    # acceptable because nothing off-host can reach it. bridged connects over 127.0.0.1, and
+    # acceptable because nothing off-host can reach it. fleetd connects over 127.0.0.1, and
     # binding 0.0.0.0 here would expose a broker with default credentials to the network.
     ports:
-      - "127.0.0.1:5672:5672"      # AMQP — bridged.yaml broker.uri points here
+      - "127.0.0.1:5672:5672"      # AMQP — fleetd.yaml broker.uri points here
       - "127.0.0.1:15672:15672"    # HTTP management API + UI
 
     # The whole point of Stage 2. Held-but-unacked replies live here; without a named volume a
@@ -57,4 +57,4 @@ services:
 
 volumes:
   lavinmq-data:
-    name: bridged-lavinmq-data
+    name: fleetd-lavinmq-data

docs/CB-201-227-Refinement.md

@@ -0,0 +1,531 @@
+# CB-201 and CB-227 refinement
+
+Date: 2026-09-03
+
+## Decision
+
+#201 and #227 are one delivery program, but they are not one implementation unit.
+
+#201 has a real seam: `CompletionResolver` can publish a typed backend-error event only after its
+waiter resolution wins. #227 can consume that event without knowing any pane text. The classifier
+must land before the final #227 wiring. However, the policy engine, roster state, and lead nudge can
+be built in parallel with the classifier.
+
+I propose five units. Units 1 to 4 own separate files and can run in parallel. Unit 5 owns all
+composition files and lands after them. It also depends on the #234 defect 2 fix named in the task.
+
+```mermaid
+flowchart LR
+    U1["Unit 1: typed backend-error classification"] --> U5["Unit 5: wire policy, spawn gate, and fleet views"]
+    U2["Unit 2: credential outage policy"] --> U5
+    U3["Unit 3: lead outage nudge"] --> U5
+    U4["Unit 4: durable member outcome"] --> U5
+    D234["#234 defect 2: fail-loud target resolution"] --> U5
+```
+
+*Figure 1. Four file-disjoint foundations feed one composition unit.*
+
+This split keeps `Fleetd.java` under one owner. It also keeps every other production file under one
+unit in this plan.
+
+## Evidence checked in the current branch
+
+I read both issue pages in full. Each page reports zero comments.
+
+| Evidence | What the code says now |
+|---|---|
+| `inject/CompletionResolver.java:229-237` | A turn below two seconds fails before normal scrape classification. A matching fast backend error is therefore only a generic failure today. |
+| `inject/CompletionResolver.java:260-276` and `:332-367` | #211 already added raw-screen classification when `lastAssistantBlock` is empty. The “dead-code question” in #201 is stale on this branch. |
+| `inject/CompletionResolver.java:288-317` | Exhaustion wins before the hard-coded `API Error:` match. A backend error then goes through generic `fail(...)`. |
+| `inject/CompletionResolver.java:311-316` | The code admits that the pattern is a heuristic. A member report which quotes an API error may match it. |
+| `inject/CompletionResolver.java:449-467` | Startup coverage exists only for `exhaustedPattern`. |
+| `inject/ExhaustedPatternLookup.java:13-25` | The current lookup and explicit `none()` value are a good shape for the new classifier seam. |
+| `Fleetd.java:196-207` | One `BackendQuarantine` is shared by placement and the exhaustion sink. Its cooldown comes from `quarantineCooldownSeconds`. |
+| `Fleetd.java:322-363` | Pattern compilation, target-to-profile lookup, and the live `ExhaustionSink` are composed in `Fleetd.main`. The sink on this branch still ends in `.ifPresent(...)`. This plan assumes #234 replaces that silent path. |
+| `placement/BackendQuarantine.java:60-87` | A repeated exhaustion restarts one long quarantine. The store is credential-keyed and uses an injected monotonic clock. |
+| `member/CompositePeerLauncher.java:260-317` | Explicit and policy-selected spawns have separate gates. Both paths must learn about outage cool-off. |
+| `member/CompositePeerLauncher.java:347-379` | Exhaustion refusal already checks a credential for explicit spawns and filters policy candidates. Its error text says “exhausted”. |
+| `placement/PlacementContext.java:10-22` and `PlacementPolicyUtil.java:14-83` | Automatic placement has only one transient exclusion set named `quarantined`. Reusing it would make outage errors say “backend exhausted”. |
+| `mcp/FleetMcp.java:913-1025` | `fleet_list` sets `free: 0` and adds `credentialId` plus `quarantinedForSeconds` when quarantine is active. |
+| `session/MemberSession.java:51-59` | The roster has `DONE` and generic `FAILED`, but no backend-error state or stored reason. |
+| `session/SessionManager.java:695-773` | A normal boundary moves `BUSY` to `DONE`. A failure moves any non-released session to `FAILED`. The async completion resolver can race the `DONE` update. |
+| `session/SessionManager.java:648-687` | `rosterView` reports the session state, but it reports no terminal reason. |
+| `msg/MessageService.java:922-940` | CB-588 already nudges for every terminal async ticket, including failures. Current code would report failed tickets, but it would not report one correlated outage. |
+| `msg/ReplyPushLoop.java:20-48` | Replies, terminal tickets, and questions share one per-lead schedule. This prevents two push sources from injecting competing turns. |
+| `msg/ReplyPushLoop.java:305-395` | Each push entry point resolves the owning lead through `PrimaryRegistry`. Missing ownership is logged and the durable or pending item remains the backstop. |
+| `msg/ReplyPushLoop.java:496-547` | One tick builds one combined nudge. Pending items have separate reminder counts. |
+| `health/FleetHealthMonitor.java:91-143` | Health is a slow periodic observer of members and message-layer facts. It does not receive completion classifications. |
+| `health/FleetHealthMonitor.java:206-208` | `healthCoverage` means health enabled plus webhook configured. It does not describe lead-pane alerts. |
+| `Fleetd.java:465-486` | Health stays `detection-only` without the webhook notification setting. |
+
+I also read the related unit tests for `CompletionResolver`, `ReplyPushLoop`, `BackendQuarantine`,
+`CompositePeerLauncher`, `PlacementPolicyUtil`, `SessionManager`, `MessageService`, and `FleetMcp`.
+
+I did not inspect the in-progress #234 branch. I only used the two measured facts in the task. No
+peer architect was named, so I did not exchange a design with one.
+
+## Required behaviour
+
+The policy should use these first values:
+
+- Threshold: **2** classified backend errors.
+- Window: **60 seconds**, measured from the first error to the second.
+- Cool-off: **60 seconds**, starting when the threshold is reached.
+- Correlation key: `credentialId`, never profile name and never error text.
+- Incident rule: one active incident per credential. Errors during its cool-off do not extend it and
+  do not create more lead notices.
+- Rearm rule: after cool-off ends, two fresh errors are needed for another incident.
+
+Two errors are the smallest threshold which protects the honest one-turn failure. A 60-second window
+fits the measured two-member outage. A 60-second cool-off blocks immediate repeat spawns without
+turning a short backend fault into the default 1,800-second exhaustion quarantine.
+
+A single classified error still fails its send and marks its member `backend_error`. It does not
+cool a credential and does not send an outage notice. This is what “a single error changes nothing”
+must mean at the credential level. It cannot mean that the failed member still looks successful.
+
+```mermaid
+sequenceDiagram
+    participant R1 as Resolver for member A
+    participant R2 as Resolver for member B
+    participant P as Outage policy
+    participant S as Spawn gate
+    participant N as Lead push loop
+    participant L as Lead pane
+
+    R1->>P: backend error for credential C
+    Note over P: Count 1, no cool-off
+    R2->>P: backend error for credential C within 60s
+    P->>P: Start one 60s incident
+    P->>S: Credential C is cooling off
+    P->>N: Queue one incident notice
+    N->>L: Inject when lead is idle, blocked, or done
+    L->>S: Request another spawn on credential C
+    S-->>L: Refuse and report remaining cool-off
+```
+
+*Figure 2. The second independent classification creates the fleet-level event.*
+
+Against the 2026-09-01 case, the second failed member would start cool-off. `fleet_list` would show
+zero free capacity and both members as `backend_error`. The push loop would inject one outage notice
+even if the lead had not polled either ticket yet. The design reports the outage. It does not recover
+uncommitted work from the members.
+
+## Unit 1 — Typed backend-error classification
+
+### Scope
+
+Replace the direct hard-coded check inside `CompletionResolver` with a lookup and a sink. Keep the
+public send result as a failed send. The typed internal event is the seam #227 consumes.
+
+The lookup returns the pattern for a target. The sink receives the target, matched line, and full
+failure reason. It fires only after `Rendezvous.resolveFailure(...)` wins for that exact captured
+waiter. This copies the race rule already used by `ExhaustionSink`.
+
+The classifier must run in all three current paths:
+
+1. a normal non-empty assistant block;
+2. the #211 raw scrape fallback;
+3. a turn inside `MIN_TURN_NANOS`, before it becomes a generic too-fast failure.
+
+In every path, the order stays: stale-baseline guard, exhaustion, backend error, then generic
+failure or completion. A fast turn still fails when no configured pattern matches.
+
+Keep `(?i)\bAPI Error\s*:` as a compatibility pattern for profiles without `errorPattern` until the
+operator config is updated. Do not call this full coverage. Startup reporting in Unit 5 must name
+profiles using this weaker legacy default.
+
+### Files owned
+
+- Add `fleetd/src/main/java/dev/ltms/fleet/inject/BackendErrorPatternLookup.java`.
+- Add `fleetd/src/main/java/dev/ltms/fleet/inject/BackendErrorSink.java`.
+- Change `fleetd/src/main/java/dev/ltms/fleet/inject/CompletionResolver.java`.
+- Change `fleetd/src/test/java/dev/ltms/fleet/inject/CompletionResolverTest.java`.
+
+No other unit may edit these files.
+
+### Acceptance criteria
+
+1. A target-specific error pattern matches a normal assistant block and resolves the send as failed.
+2. The same match calls `BackendErrorSink` exactly once after the waiter resolution wins.
+3. A late classification which loses to `fleet_reply` does not call the sink.
+4. An exhausted line that also matches the generic error pattern stays `BACKEND_EXHAUSTED`. It calls
+   only `ExhaustionSink`.
+5. A raw pane with leading Terminal User Interface (TUI) chrome and no assistant marker still uses
+   the #211 fallback and calls the backend-error sink.
+6. A matching error inside the two-second floor is typed and sent to the sink. A non-matching fast
+   turn stays a generic failure.
+7. An unchanged delivery baseline which contains old backend-error text is suppressed. It never
+   increments outage evidence.
+8. A non-match keeps the existing completion result and text.
+9. Constructors used by current callers keep compiling. They use the legacy default lookup and an
+   explicit inert sink until Unit 5 supplies the production objects.
+10. Unit tests pass. The developer runs the focused test first, then `mvn clean install` from
+    `fleetd/`.
+
+### Dependencies
+
+None. Unit 1 can run with Units 2, 3, and 4.
+
+Unit 5 depends on its new lookup, sink, and constructor.
+
+### What to report back
+
+- The exact classifier order in all three paths.
+- The focused test command and result.
+- The test which proves a losing waiter race does not publish an event.
+- The test which proves a fast matching failure is typed.
+- The final `mvn clean install` result.
+- Any constructor kept only for transition and where Unit 5 replaces it.
+
+## Unit 2 — Credential outage policy
+
+### Scope
+
+Build a small credential-keyed state machine. It accepts already-classified backend-error events.
+It does not read pane text, profiles, sessions, or lead state.
+
+Use an injected monotonic clock. A call records `credentialId`, target, and reason. It returns a new
+incident only on the threshold crossing. The incident contains a stable event id, credential id,
+the distinct affected targets, evidence count, window, and remaining cool-off.
+
+This class owns both correlation and short cool-off. Keeping them together makes threshold crossing
+and the cool-off deadline one atomic state change.
+
+### Files owned
+
+- Add `fleetd/src/main/java/dev/ltms/fleet/placement/BackendOutagePolicy.java`.
+- Add `fleetd/src/test/java/dev/ltms/fleet/placement/BackendOutagePolicyTest.java`.
+
+No other unit may edit these files.
+
+### Acceptance criteria
+
+1. One error creates no incident and no cool-off.
+2. Two errors for one credential within 60 seconds create exactly one incident and a 60-second
+   cool-off.
+3. Two errors more than 60 seconds apart do not create an incident.
+4. The exact 60-second boundary has a pinned result. Use inclusive `<= 60s` so scheduler delay does
+   not discard evidence at the boundary.
+5. Different credentials never share evidence.
+6. Different profiles which supply the same credential id do share evidence. The policy itself only
+   sees the credential id.
+7. More errors during active cool-off do not extend its deadline and do not return another incident.
+8. After expiry, old evidence is cleared. Two fresh errors are needed to create the next incident.
+9. Remaining seconds round up, matching `BackendQuarantine` reporting.
+10. Concurrent second and third errors cannot return two incidents.
+11. The focused tests and `mvn clean install` pass.
+
+### Dependencies
+
+None. Unit 2 can run with Units 1, 3, and 4.
+
+Unit 5 depends on the policy API.
+
+### What to report back
+
+- The state transition table and locking method.
+- The exact threshold, window, cool-off, and boundary rule.
+- The test which proves one incident under concurrent calls.
+- The focused test and `mvn clean install` results.
+
+## Unit 3 — Lead outage nudge
+
+### Scope
+
+Add backend incidents as a fourth pending source in `ReplyPushLoop`. Do not create another scheduler
+or call `AgentControl.send` from `Fleetd`. The existing combined per-lead schedule is the control
+which prevents competing injected turns.
+
+The entry point takes an incident id, affected worker targets, credential id, affected profile
+names, and remaining cool-off. It resolves distinct owning leads through `PrimaryRegistry`.
+
+Each `(incidentId, lead)` item is one-shot. It waits while the lead is not injectable. After one
+successful `agents.send`, remove it. A send exception keeps it pending for a bounded retry. It never
+uses the repeated reminder behaviour of an uncollected ticket.
+
+Also add a fail-loud entry point for a classified target that Unit 5 cannot map to a credential. It
+uses `PrimaryRegistry.nudgeTargetFor(target)` and says that correlation could not run. If no lead is
+known, log at `WARN`, not `DEBUG`.
+
+### Files owned
+
+- Change `fleetd/src/main/java/dev/ltms/fleet/msg/ReplyPushLoop.java`.
+- Change `fleetd/src/test/java/dev/ltms/fleet/msg/ReplyPushLoopTest.java`.
+
+No other unit may edit these files.
+
+### Acceptance criteria
+
+1. One incident affecting two workers owned by one lead causes one successful pane injection.
+2. Two affected workers owned by two leads cause one successful injection per affected lead. This
+   is one notice per event per lead, not one notice per member.
+3. Repeating the same incident id is idempotent.
+4. A busy or unknown lead is not injected. The item stays pending until the lead becomes injectable
+   or its attempt cap is reached.
+5. After one successful injection, later ticks do not mention that incident again.
+6. A failed `agents.send` is retried within the existing bound. A successful retry still gives only
+   one successful send.
+7. A pending ticket and an outage incident for one lead appear in one combined nudge, not two
+   competing turns.
+8. The text names the credential, profiles, affected workers, and remaining cool-off. It tells the
+   lead to run `fleet_list`.
+9. An unmapped target produces a direct warning notice when a lead is known. If no lead is known,
+   the code logs a `WARN` naming the target and reason.
+10. `stop()` clears incident state as it clears other push state.
+11. Existing reply, ticket, and question tests stay green. The focused tests and
+    `mvn clean install` pass.
+
+### Dependencies
+
+None. The API uses plain values, not the Unit 2 incident class. This lets Unit 3 run in parallel.
+
+Unit 5 adapts the Unit 2 incident into this entry point.
+
+### What to report back
+
+- The exact one-shot and retry rules.
+- The test showing one combined nudge with a failed ticket.
+- The test showing one successful send for two affected workers.
+- The focused test and `mvn clean install` results.
+
+## Unit 4 — Durable member backend-error outcome
+
+### Scope
+
+Make a classified backend failure remain visible after its ticket is collected or expires.
+
+Add `BACKEND_ERROR` to `MemberSession.State`. Add a nullable failure detail to `MemberSession` and
+render it as `failureReason` in `SessionManager.rosterView`. Add
+`SessionManager.onBackendError(target, reason)`.
+
+The transition must handle both completion orderings:
+
+- `BUSY -> BACKEND_ERROR` when classification wins before the normal completion state update;
+- `DONE -> BACKEND_ERROR` when the async resolver runs after `SessionManager.onTurnComplete`.
+
+It must use a compare-and-set retry or another atomic update. `RELEASED` must never return to the
+roster. A backend-error member is terminal and cannot accept another delivery.
+
+### Files owned
+
+- Change `fleetd/src/main/java/dev/ltms/fleet/session/MemberSession.java`.
+- Change `fleetd/src/main/java/dev/ltms/fleet/session/SessionManager.java`.
+- Change `fleetd/src/test/java/dev/ltms/fleet/session/SessionManagerTest.java`.
+
+No other unit may edit these files.
+
+### Acceptance criteria
+
+1. `onBackendError` moves a `BUSY` member to `BACKEND_ERROR` and stores the reason.
+2. It also moves `DONE` to `BACKEND_ERROR`, covering the resolver race.
+3. A later normal `onTurnComplete` cannot change `BACKEND_ERROR` back to `DONE`.
+4. A released or unknown member is not recreated. The unknown case logs at `WARN` and returns an
+   explicit false result to its caller.
+5. `onDelivered` refuses a `BACKEND_ERROR` member, just as it refuses generic `FAILED`.
+6. `rosterView` reports `state: backend_error` and `failureReason` after the send ticket is gone.
+7. Ordinary members do not gain a blank or invented `failureReason` field.
+8. Existing constructors keep source compatibility for tests and adapters.
+9. The focused tests and `mvn clean install` pass.
+
+### Dependencies
+
+None. Unit 4 can run with Units 1, 2, and 3.
+
+Unit 5 calls the new session method from the production sink.
+
+### What to report back
+
+- The two race orderings and the tests for both.
+- The exact roster JSON shape.
+- The unknown-target result and log level.
+- The focused test and `mvn clean install` results.
+
+## Unit 5 — Production wiring, spawn gate, and fleet views
+
+### Scope
+
+Compose Units 1 to 4 in production. This is the only unit which edits `Fleetd.java`.
+
+Add per-profile `errorPattern` config beside `exhaustedPattern`. Compile both once at startup. A
+configured pattern wins over the legacy default. Report configured profiles and legacy-default
+profiles separately at startup. A bad regex must stop startup with the profile and key in the
+message.
+
+Wire one production `BackendErrorSink` with this order:
+
+1. mark the member `backend_error` with its reason;
+2. resolve the profile and its current `effectiveCredentialId()` through the fail-loud #234 seam;
+3. record the error in `BackendOutagePolicy`;
+4. on a new incident, submit one event to `ReplyPushLoop`.
+
+If target metadata cannot be resolved, do not end in `Optional.ifPresent`. Log an error and call the
+Unit 3 unmapped-target notice. The failed send still reaches its ticket through CB-588.
+
+Teach both spawn paths about a separate cool-off source. Exhaustion quarantine has priority when
+both states are active. Automatic placement needs a distinct `coolingOff` set so its refusal does
+not say “exhausted”.
+
+Extend the MCP (Model Context Protocol) views:
+
+- A cooling profile has `free: 0`, `credentialId`, and `coolingOffForSeconds` in `fleet_list`.
+- It does not have `quarantinedForSeconds` unless exhaustion quarantine is also active.
+- `fleet_profiles` has a separate `coolingOff` map, not an entry in `quarantined`.
+- A direct spawn refusal says the credential is cooling off after repeated backend errors and gives
+  the remaining seconds.
+
+Do not change `FleetHealthMonitor.coverage`. It still describes the periodic health webhook path.
+Lead-pane outage delivery is a separate capability.
+
+### Files owned
+
+- `fleetd/src/main/java/dev/ltms/fleet/Fleetd.java`
+- `fleetd/src/main/java/dev/ltms/fleet/config/FleetConfig.java`
+- `fleetd/src/main/java/dev/ltms/fleet/config/ConfigRef.java`
+- `fleetd/src/main/java/dev/ltms/fleet/member/CompositePeerLauncher.java`
+- `fleetd/src/main/java/dev/ltms/fleet/placement/PlacementContext.java`
+- `fleetd/src/main/java/dev/ltms/fleet/placement/PlacementPolicyUtil.java`
+- `fleetd/src/main/java/dev/ltms/fleet/mcp/FleetMcp.java`
+- `fleetd/src/test/java/dev/ltms/fleet/config/FleetConfigTest.java`
+- `fleetd/src/test/java/dev/ltms/fleet/config/ConfigRefTest.java`
+- `fleetd/src/test/java/dev/ltms/fleet/member/CompositePeerLauncherTest.java`
+- `fleetd/src/test/java/dev/ltms/fleet/placement/PlacementPolicyTest.java`
+- `fleetd/src/test/java/dev/ltms/fleet/mcp/FleetMcpTest.java`
+- Add `fleetd/src/test/java/dev/ltms/fleet/BackendOutageFlowTest.java`.
+- `fleetd/fleetd.example.yaml`
+- `CLAUDE.md`
+
+No earlier unit edits these files.
+
+The lead, not a worker, must update `wiki/7-Use-Cases.md`, `wiki/9-Implementation.md`, and
+`wiki/11-Features.md`. Project rules forbid workers from committing `wiki/`. The portable block in
+`CLAUDE.md` and `wiki/7-Use-Cases.md` must remain byte-identical.
+
+### Acceptance criteria
+
+1. `errorPattern` binds per profile. Blank uses the legacy default and is reported as degraded
+   coverage. A config reload which changes it is reported as deferred because patterns are compiled
+   at startup.
+2. A malformed `errorPattern` stops startup and names `profiles.<name>.errorPattern`.
+3. The real production sink never silently drops an unknown target. A test captures its error log
+   and the fallback notice call.
+4. One real backend-error classification through `CompletionResolver` marks only its member. It does
+   not cool the credential and does not send an outage notice.
+5. Two real classifications for one credential within 60 seconds start one incident.
+6. The integration test then calls the real explicit-profile spawn gate. It is refused before any
+   adapter spawn call, with “cooling off” and remaining seconds in the message.
+7. The same test calls an automatic placement path. A cooling candidate is skipped. If every
+   candidate is cooling, the error names cool-off rather than exhaustion.
+8. Profiles sharing the credential are all blocked. A profile on another credential stays usable.
+9. `fleet_list` from the same fixture shows both members as `backend_error`, preserves each failure
+   reason, and reports `free: 0`, the credential, and `coolingOffForSeconds`.
+10. `fleet_profiles` reports cool-off separately from quarantine.
+11. The real `ReplyPushLoop` receives one incident and makes one successful lead-pane send. Existing
+    failed-ticket notice content may share that same combined send.
+12. Exhaustion still wins when a line matches both patterns. A simultaneous exhaustion quarantine
+    also wins in spawn errors and fleet views.
+13. After the 60-second cool-off, spawn is allowed again. A new incident needs two fresh errors.
+14. `healthCoverage` has the same value before and after this change for the same health config.
+15. `fleetd.example.yaml` explains `errorPattern`, the legacy fallback, 2/60/60 policy, and the
+    difference between cool-off and exhaustion quarantine.
+16. `CLAUDE.md` tells leads how `fleet_profiles` and `fleet_list` report cool-off. The lead later
+    applies the matching wiki updates and runs the documented byte-sync check.
+17. The developer records the new end-to-end test failing before implementation, then passing. The
+    focused suites and `mvn clean install` pass.
+
+### Dependencies
+
+Unit 5 starts only after Units 1 to 4 are merged or rebased into its branch. It also starts after the
+#234 defect 2 fix lands, because both areas touch the same target-resolution control path.
+
+### What to report back
+
+- The exact commits used for Units 1 to 4 and #234.
+- The startup coverage line with one configured and one legacy-default profile.
+- The red test output before the implementation and its green result after.
+- The explicit and automatic spawn refusal text.
+- Sample `fleet_list` and `fleet_profiles` JSON for cool-off and exhaustion.
+- The number and text of lead-pane sends in the real-path test.
+- The focused test commands and final `mvn clean install` result.
+- The exact `CLAUDE.md` change and the wiki edits the lead must apply.
+
+## File ownership summary
+
+| Area | Unit | Shared edit risk |
+|---|---:|---|
+| Completion classification | 1 | Only Unit 1 edits `CompletionResolver` and its test. |
+| Correlation and cool-off state | 2 | New files only. |
+| Lead push scheduling | 3 | Only Unit 3 edits `ReplyPushLoop` and its test. |
+| Member terminal state | 4 | Only Unit 4 edits `MemberSession`, `SessionManager`, and their test. |
+| Main composition, config, placement, MCP views, shipped prompt | 5 | Only Unit 5 edits `Fleetd`, `FleetConfig`, `CompositePeerLauncher`, placement context, `FleetMcp`, and `CLAUDE.md`. |
+| Wiki propagation | Lead after Unit 5 | Workers do not commit the wiki submodule. |
+
+## What I would not build
+
+1. **Do not reuse `BackendQuarantine` for outages.** Its repeat call restarts a long credential
+   quarantine. Its fields and errors say “exhausted”. That is wrong for a short outage.
+2. **Do not merge the exhaustion and generic error patterns.** Exhaustion must win because it has a
+   different policy and duration.
+3. **Do not group by error string.** One outage can produce different text. The shared operational
+   limit is the credential.
+4. **Do not mark a profile unusable until config changes.** The current classifier cannot safely
+   tell a permanent malformed request from a transient service fault. A permanent state would need
+   a stronger error taxonomy first.
+5. **Do not quarantine on the first generic backend error.** That would turn one bad request or one
+   false pattern match into a fleet-wide capacity loss.
+6. **Do not add this to `FleetHealthMonitor`.** The monitor samples slow member health. The exact
+   backend event already exists at completion resolution, and moving it to polling would lose type
+   and time.
+7. **Do not add another direct lead injector.** `ReplyPushLoop` already owns status gating,
+   per-lead coalescing, retry bounds, and heartbeat stand-down.
+8. **Do not change `healthCoverage` to `full`.** That field still means a webhook notification sink
+   exists for periodic health. A backend outage nudge does not make every health event visible.
+9. **Do not persist incident history across daemon restart in this work.** Existing exhaustion
+   quarantine is also in memory. A 60-second state does not justify a new durable store.
+10. **Do not build work recovery.** The PR-body survival story proves why checkpoint-first work is
+    useful, but these tickets are about detection, capacity, and signalling.
+11. **Do not remove the legacy `API Error:` fallback in the first release.** Doing so would turn an
+    unedited config back into a false successful completion. Report it as degraded coverage instead.
+12. **Do not reorder or add the old `visibleTurn` fallback from #201.** #211 already implemented the
+    narrow raw-scrape fallback at `CompletionResolver.classifyRawScrapeFallback`.
+
+## Riskiest assumption and cheapest experiment
+
+The riskiest assumption is that a configured error regex means “the backend failed this turn”. The
+current code and test already show the counterexample: a worker may quote `API Error:` while writing
+a valid report. Two such false matches on one credential would now remove capacity for 60 seconds.
+
+The cheapest experiment is a replay corpus before Unit 5 ships:
+
+1. Save the full pane text from the measured 2026-09-01 outage.
+2. Produce one safe failure per backend with a disposable invalid endpoint or request.
+3. Save one valid member report which quotes each error line.
+4. Replay all samples through the real `CompletionResolver` test fixture.
+5. Require outage samples to match and quoted-report samples not to match after assistant-block
+   extraction and baseline checks.
+
+This costs no outage deployment and no real sleep. If quoted reports still match, narrow the profile
+patterns before enabling correlation. Do not raise the threshold to hide a bad classifier.
+
+## Sequencing with three developers
+
+First wave:
+
+1. Developer A: Unit 1, typed classification.
+2. Developer B: Unit 2, credential outage policy.
+3. Developer C: Unit 3, lead outage nudge.
+
+As soon as one slot is free, start Unit 4. It is file-disjoint from every first-wave unit. Merge and
+review Units 1 to 4 independently.
+
+Start Unit 5 only after all four foundations and #234 are available. Unit 5 is the only high-conflict
+integration branch, so no other active unit should touch its file list.
+
+## Checks performed for this refinement
+
+- Read issue #201 and issue #227 through their Gitea pages. Both showed zero comments.
+- Read the source and tests named in the evidence section.
+- Ran `git status --short --branch`; the branch was clean before this document was added.
+- Ran `git log --oneline -12` to identify the branch base.
+- I did not run Maven because this change adds only a design document.
+- Rendered both Mermaid blocks with `npx @mermaid-js/mermaid-cli`; both commands succeeded.

docs/CB-301-Session-Manager.md

@@ -2,7 +2,7 @@
 
 **Status:** design spec for review → delegate implementation.
 **Grounded in:** `WorkerService`, `Injector`/`StatusPoller`/`TurnListener`, `MessageService`,
-`BridgeMcp`, `BridgedApp` (see [wiki 9. Implementation](../wiki/9-Implementation.md)).
+`FleetMcp`, `FleetApp` (see [wiki 9. Implementation](../wiki/9-Implementation.md)).
 
 ## Problem
 
@@ -15,7 +15,7 @@ Consequences today:
   since when?" without shelling to herdr for a raw agent list (no state, no ownership, no age).
 - Cleanup of a worker that outlived its owning process depends entirely on the boot-time
   name-nonce **reaper** (CB-117) — there is no live, authoritative roster during a run.
-- `bridge_list` (CB-304) can only surface herdr's view, not a bridge-owned roster.
+- `fleet_list` (CB-304) can only surface herdr's view, not a bridge-owned roster.
 - There is no seam for per-session policy (checkpoint on teardown → CB-302; idle_ttl /
   context_cap / drain → CB-303).
 
@@ -43,7 +43,7 @@ Under no-reuse, a released session is terminal. A new `acquire` always creates a
 subscription-guarded spawn/teardown mechanics; `SessionManager` adds the registry, lifecycle, and
 ownership on top.
 
-**Package:** new `dev.ltms.bridged.session` — keeps the registry/lifecycle concern separate from
+**Package:** new `dev.ltms.fleet.session` — keeps the registry/lifecycle concern separate from
 the `worker` spawn mechanics. Holds `SessionManager` + `WorkerSession`.
 
 ### `WorkerSession` (record or small mutable holder)
@@ -96,13 +96,13 @@ final class SessionManager {
 
 ### Integration points
 
-- **`Bridged.main`** — construct `SessionManager(workerService, ...)`; wire it as/decorating the
+- **`Fleetd.main`** — construct `SessionManager(workerService, ...)`; wire it as/decorating the
   `TurnListener` alongside `CompletionResolver` so it sees turn boundaries, and give it the
   `WorkerPresence` signal for `READY`.
-- **`BridgeMcp.spawn` / `BridgedApp.spawnWorker`** — route spawn through `SessionManager.acquire`
-  (carry `callerTerminal` as `ownerTerminal`). **`bridge_stop` / `DELETE /workers/{paneId}`** →
+- **`FleetMcp.spawn` / `FleetApp.spawnWorker`** — route spawn through `SessionManager.acquire`
+  (carry `callerTerminal` as `ownerTerminal`). **`fleet_stop` / `DELETE /workers/{paneId}`** →
   `SessionManager.release`.
-- **`bridge_list` / `GET /sessions` (CB-304 later)** — read `SessionManager.roster()`.
+- **`fleet_list` / `GET /sessions` (CB-304 later)** — read `SessionManager.roster()`.
 - **`MessageService`** — no change required for one-shot; a later CB-303 auto-release hook can call
   `release` from `onTurnComplete` under policy.
 
@@ -121,4 +121,4 @@ final class SessionManager {
 - **CB-302** — attach a checkpoint step (`STATE.md` + commit) to the `release` path.
 - **CB-303** — a policy loop over `roster()` using `spawnedAtNanos`/state to auto-`release` on
   `idle_ttl`, or drain on `context_cap`.
-- **CB-304** — `bridge_list` reads `roster()` for a bridge-owned roster + live join.
+- **CB-304** — `fleet_list` reads `roster()` for a bridge-owned roster + live join.

docs/CB-301-ext-Worktree-Provisioning.md

@@ -2,12 +2,12 @@
 
 **Status:** ✅ shipped — implemented at commit `97ecc71` (per-worker git worktree + config-parity
 overlay). As-built: `session/GitWorktrees.java` behind the `Worktrees` port, wired in
-`Bridged.main` and configurable via `worktreeRoot` / per-profile `parityOverlay`
-(see `bridged.example.yaml`). Branch/worktree surface in `bridge_list` landed with CB-304
+`Fleetd.main` and configurable via `worktreeRoot` / per-profile `parityOverlay`
+(see `fleetd.example.yaml`). Branch/worktree surface in `fleet_list` landed with CB-304
 (`9fe04bf`); the worker-opened-PR checkpoint landed as CB-302 (`64e70ef`).
 **Extends:** [CB-301 Session Manager](CB-301-Session-Manager.md) (shipped, commit `54d907c`).
 **Realizes:** the config-parity requirement in [Worker Git Workflow](Worker-Git-Workflow.md).
-**Grounded in:** `SessionManager`, `WorkerService.spawn/effectiveCwd`, `BridgedConfig.Worker`,
+**Grounded in:** `SessionManager`, `WorkerService.spawn/effectiveCwd`, `FleetConfig.Worker`,
 `inject/…LsofPeerPidLookup` (the `ProcessBuilder` exec pattern).
 
 ## Problem
@@ -43,7 +43,7 @@ provider.
 ### `WorktreeRequest` (new, nullable = "no worktree")
 
 ```java
-package dev.ltms.bridged.session;
+package dev.ltms.fleet.session;
 /** Ask acquire() to provision an isolated worktree. null ⇒ run in the shared primary tree. */
 public record WorktreeRequest(String ticketSlug, String baseRef) {
     // ticketSlug seeds the branch name; baseRef null/blank ⇒ current HEAD of the repo.
@@ -63,7 +63,7 @@ untouched.
 ### `Worktrees` seam (new)
 
 ```java
-package dev.ltms.bridged.session;
+package dev.ltms.fleet.session;
 public interface Worktrees {
     /** git -C <repoRoot> worktree add <path> -b <branch> <baseRef|HEAD>. Returns the worktree path. */
     String add(String repoRoot, String branch, String baseRef);
@@ -112,7 +112,7 @@ public void release(String paneId) {
 }
 ```
 
-### Config — `BridgedConfig.Worker.parityOverlay` + a `worktreeRoot`
+### Config — `FleetConfig.Worker.parityOverlay` + a `worktreeRoot`
 
 - Add `List<String> parityOverlay` to the `Worker` record (12th field). Compact-constructor default
   when null/empty: `[".mcp.json", ".claude/settings.local.json", ".env", ".envrc"]` (missing paths are
@@ -122,7 +122,7 @@ public void release(String paneId) {
 
 ### Surface: MCP + REST
 
-- `bridge_spawn` gains an optional `worktree` arg: `true`, or a ticket slug string. Truthy ⇒ build a
+- `fleet_spawn` gains an optional `worktree` arg: `true`, or a ticket slug string. Truthy ⇒ build a
   `WorktreeRequest(slug, null)` and call the 5-arg `acquire`.
 - `POST /workers` gains `worktree` (+ optional `ticket`) in the body/query, same mapping.
 - `workerView`/`view(WorkerSession)` include `worktree` and `branch` **when non-null** (omit for

docs/CB-306-Spawn-Readiness-Gate.md

@@ -1,16 +1,16 @@
 # CB-306 — Spawn-Readiness Gate (launcher-owned terminal readiness)
 
 **Status:** design note / delegation spec (branch `worker/cb-306-readiness`)
-**Issue:** gitea `lms/claude-bridge` #4
+**Issue:** gitea `fleet/fleetd` #4
 **Owner of the behaviour:** `ClaudeCodeLauncher` (the `PeerLauncher` adapter) — NOT core.
 
 ## 1. Problem
 
-`bridge_spawn` today returns a session the instant the herdr pane is started. The pane is not
+`fleet_spawn` today returns a session the instant the herdr pane is started. The pane is not
 yet a usable Claude REPL — it may still be sitting at the folder-trust prompt, or the CLI may
 never come up at all. Nothing blocks or times out on that. Consequences:
 
-- A `bridge_send` to a not-yet-ready worker surfaces as a **~60 s MCP-client timeout** (the send
+- A `fleet_send` to a not-yet-ready worker surfaces as a **~60 s MCP-client timeout** (the send
   blocks waiting for a turn that can't start) instead of a fast, explicit spawn failure.
 - A worker stuck at the folder-trust prompt lingers in `SPAWNING` forever; nothing fails it.
 
@@ -55,7 +55,7 @@ a crash, and a slow start all present as "never becomes injectable" and all corr
    or `spawnReadyTimeoutMs` elapses.
 3. **Ready** → return the `WorkerHandle(paneId, terminalId)` as today.
 4. **Timeout** → the launcher **closes the pane it started** (and its tab, via the same path
-   `release`/`stop` uses) and throws **`PeerUnreachableException`** (new, in `dev.ltms.bridged.peer`).
+   `release`/`stop` uses) and throws **`PeerUnreachableException`** (new, in `dev.ltms.fleet.peer`).
    No orphan pane is left behind — the launcher cleans up its own failed birth.
 
 `spawnReadyTimeoutMs == 0` (or unset) **disables** the gate = legacy non-blocking behaviour, so the
@@ -79,8 +79,8 @@ Unit tests (add to the existing `ClaudeCodeLauncher` test):
 
 ## 5. Config
 
-Add to the launcher-level config (a bridged-level knob, not per-profile) in `bridged.yaml` +
-`BridgedConfig`:
+Add to the launcher-level config (a fleetd-level knob, not per-profile) in `fleetd.yaml` +
+`FleetConfig`:
 
 ```yaml
 spawn_ready_timeout_ms: 20000   # 0 disables the gate (legacy non-blocking spawn)
@@ -88,7 +88,7 @@ spawn_ready_poll_ms: 300
 ```
 
 Jackson ignores unknown keys, so omitting them in existing YAML is safe; pick sane defaults in code
-(`20000` / `300`). Keep the names consistent with existing config field style in `BridgedConfig`.
+(`20000` / `300`). Keep the names consistent with existing config field style in `FleetConfig`.
 
 ## 6. Core / MCP propagation
 
@@ -99,11 +99,11 @@ Jackson ignores unknown keys, so omitting them in existing YAML is safe; pick sa
   `spawn` throws — verify the new exception flows through it (worktree removed, nothing registered).
 - The **non-worktree** path registers the session only *after* `spawn` returns, so a throw means no
   half-live `SPAWNING` session is ever registered — confirm this and add a test.
-- `bridge_spawn` (MCP verb) must return an **error result** carrying the exception message, not a
-  success with a dead session. Trace `BridgeMcp`/`BridgedApp` spawn handlers and make sure the
+- `fleet_spawn` (MCP verb) must return an **error result** carrying the exception message, not a
+  success with a dead session. Trace `FleetMcp`/`FleetApp` spawn handlers and make sure the
   exception becomes a clean tool error, not an uncaught 500 with a stack trace.
 
-**Out of scope (do NOT do here):** gating `bridge_send` on session `READY` (existing status-gate +
+**Out of scope (do NOT do here):** gating `fleet_send` on session `READY` (existing status-gate +
 this spawn gate already close the window), MCP-handshake-as-readiness signal, the CB-307 broker,
 any config `kind:` discriminator, any second adapter.
 
@@ -111,7 +111,7 @@ any config `kind:` discriminator, any second adapter.
 
 - `ClaudeCodeLauncher.spawn` blocks until injectable or throws `PeerUnreachableException` +
   self-reaps the pane; gate disabled when timeout is 0.
-- New `PeerUnreachableException` in `dev.ltms.bridged.peer`.
+- New `PeerUnreachableException` in `dev.ltms.fleet.peer`.
 - Config knobs wired (`spawn_ready_timeout_ms`, `spawn_ready_poll_ms`) with safe defaults.
 - Existing `SPAWNING→READY` MCP-contact transition untouched.
 - New unit tests (ready / timeout+reap / disabled) green; **all existing tests still pass unchanged**.

docs/CB-307-Reliable-Delivery.md

@@ -13,28 +13,28 @@ Stage 2 (the AMQP/LavinMQ adapter behind the same port) is explicitly **out of s
 ## 1. The bug this fixes (grounded in current code)
 
 The reverse (worker→primary) path is `Rendezvous` — a `ConcurrentHashMap<session, CompletableFuture<Resolution>>`
-of **live blocking waiters only**. No queue, no store. When a worker calls `bridge_reply` and **no send
+of **live blocking waiters only**. No queue, no store. When a worker calls `fleet_reply` and **no send
 is currently open** for that worker:
 
 - `Rendezvous.resolve(session, content)` → `complete(...)` → `waiters.get(session) == null` →
   returns `false` (`msg/Rendezvous.java:212-215`).
 - The `content` string is **never retained** — it is dropped. The worker is told it failed:
-  `BridgeMcp.reply` returns `error("no send is awaiting a reply for this worker")` (`mcp/BridgeMcp.java:270-272`);
-  REST returns `409 no_pending_send` (`rest/BridgedApp.java:339-345`).
+  `FleetMcp.reply` returns `error("no send is awaiting a reply for this worker")` (`mcp/FleetMcp.java:270-272`);
+  REST returns `409 no_pending_send` (`rest/FleetApp.java:339-345`).
 
-This is the observed "communication break": a worker that finishes just after its `bridge_send` timed
+This is the observed "communication break": a worker that finishes just after its `fleet_send` timed
 out (the ~60s sync window) replies into the void. There is **no message-id, dedup, or ack** anywhere in
 the message path today.
 
 ## 2. What to build
 
-### 2.1 The port — `dev.ltms.bridged.msg.ReplyInbox`
+### 2.1 The port — `dev.ltms.fleet.msg.ReplyInbox`
 
 A thin interface owned by the `msg` layer. The in-memory adapter is Stage 1; the AMQP adapter (Stage 2)
 implements the **same** interface, so keep it broker-agnostic.
 
 ```java
-package dev.ltms.bridged.msg;
+package dev.ltms.fleet.msg;
 
 import java.util.List;
 
@@ -70,7 +70,7 @@ public interface ReplyInbox {
   seen-set — your call; preserve insertion order).
 - `peek` returns an immutable copy; `ack` removes by `msgId`. Thread-safe (concurrent publish vs. drain).
 - **This is soft-state, NOT persistence.** Lost on a `java -jar` bounce — that is correct and consistent
-  with "bridged stays soft-state." Do **not** add any file/DB backing.
+  with "fleetd stays soft-state." Do **not** add any file/DB backing.
 
 ### 2.3 Publish seam — route reply through the service layer
 
@@ -79,7 +79,7 @@ in `MessageService`, which already owns the `Rendezvous` and will own the `Reply
 
 - Add `MessageService.reply(String session, String content)`:
   ```java
-  /** Route a worker's explicit bridge_reply: resolve an open send, or queue it in the inbox if none. */
+  /** Route a worker's explicit fleet_reply: resolve an open send, or queue it in the inbox if none. */
   public boolean reply(String session, String content) {
       if (rendezvous.resolve(session, content)) {
           return true;                       // a live send took it — unchanged fast path
@@ -89,12 +89,12 @@ in `MessageService`, which already owns the `Rendezvous` and will own the `Reply
   }
   ```
 - Repoint the two callers off the bare `rendezvous.resolve(...)` onto `messages.reply(...)`:
-  - `BridgeMcp.reply` (`mcp/BridgeMcp.java:262-273`) — on success return a normal ack; **remove** the
+  - `FleetMcp.reply` (`mcp/FleetMcp.java:262-273`) — on success return a normal ack; **remove** the
     `error("no send is awaiting a reply…")` branch (that case is now a successful queue).
-  - `BridgedApp.replyMessage` (`rest/BridgedApp.java:330-346`) — return `200` (queued) instead of
+  - `FleetApp.replyMessage` (`rest/FleetApp.java:330-346`) — return `200` (queued) instead of
     `409 no_pending_send`.
 
-**DO NOT touch the QUESTION path.** `bridge_ask` / `rendezvous.resolveQuestion` must keep today's
+**DO NOT touch the QUESTION path.** `fleet_ask` / `rendezvous.resolveQuestion` must keep today's
 `NO_WAITER` behaviour — a mid-turn question is **interactive** (the worker blocks synchronously and cannot
 consume a late answer), so it must **never** be queued. Only terminal `REPLY`s go to the inbox.
 
@@ -110,27 +110,27 @@ The primary re-checks a worker it delegated to. Expose a drain keyed by **worker
 - Add `MessageService.drainReplies(String target)`: `peek` the inbox, `ack` each returned `msgId`, hand
   back the `List<InboxMessage>` (or just the contents). At-least-once: peek→deliver→ack (ack only after
   the caller has them, so an in-flight failure re-surfaces them).
-- **DECISION (required default): expose via the existing poll verb, keyed by target.** Extend `bridge_poll`
+- **DECISION (required default): expose via the existing poll verb, keyed by target.** Extend `fleet_poll`
   to accept an optional `target` (worker session) and, when present, return that worker's drained replies —
   alongside a matching REST route `GET /sessions/{id}/replies`. Do **not** change `send`/`answer` semantics
   (do not drain inside `send` — that conflates "deliver to worker" with "collect its mail"). Keep the
-  existing ticket-based `bridge_poll(ticket)` path working unchanged. If you see a cleaner surface, still
+  existing ticket-based `fleet_poll(ticket)` path working unchanged. If you see a cleaner surface, still
   ship this default and note the alternative for review.
 
 ## 3. Config
 
 **None for Stage 1.** The in-memory adapter is the unconditional default — wire `new InMemoryReplyInbox()`
-into `MessageService` in `Bridged.main`. Do **not** add a `broker:` config block (that arrives with the
+into `MessageService` in `Fleetd.main`. Do **not** add a `broker:` config block (that arrives with the
 Stage-2 AMQP adapter: absent → in-memory, present → AMQP).
 
 ## 4. Acceptance criteria (what the primary will verify)
 
-1. New `ReplyInbox` + `InboxMessage` + `InMemoryReplyInbox` in `dev.ltms.bridged.msg`.
-2. `bridge_reply` with **no open send** now **succeeds and queues** (no more `error` / `409`); the reply is
+1. New `ReplyInbox` + `InboxMessage` + `InMemoryReplyInbox` in `dev.ltms.fleet.msg`.
+2. `fleet_reply` with **no open send** now **succeeds and queues** (no more `error` / `409`); the reply is
    later retrievable and identical.
 3. The queued reply is drainable by the primary keyed by target; draining **acks** it (a second drain
    returns nothing); dedup by `msgId` (re-publishing the same id does not double-queue).
-4. **QUESTION path unchanged** — `bridge_ask` with no open send still returns `NO_WAITER` (add/keep a test
+4. **QUESTION path unchanged** — `fleet_ask` with no open send still returns `NO_WAITER` (add/keep a test
    proving a question is never queued).
 5. Completion/failure fallbacks unchanged.
 6. Unit tests covering: `InMemoryReplyInbox` publish/peek/ack/dedup/FIFO/concurrency; `MessageService.reply`
@@ -140,7 +140,7 @@ Stage-2 AMQP adapter: absent → in-memory, present → AMQP).
 
 ## 5. Build & verification (worker side)
 
-- Build with Maven from the worktree's `bridged/` dir. **Capture the exit code without a masking pipe**
+- Build with Maven from the worktree's `fleetd/` dir. **Capture the exit code without a masking pipe**
   (`mvn clean install; echo "MVN_EXIT=$?"` — never `mvn … | tail`, which hides failures).
 - Read the real test totals from `target/surefire-reports/TEST-*.xml`, not from stdout scroll.
 - You do **not** have IDE MCP access — do not claim `ide_diagnostics` results. The **primary** runs the
@@ -152,7 +152,7 @@ Stage-2 AMQP adapter: absent → in-memory, present → AMQP).
 - **`.mcp.json` is `--skip-worktree` in your worktree — never edit, `git add`, or commit it.**
 - **`wiki/` is a submodule — never run git in it; never touch it.**
 - Commit only your feature changes (the new port/adapter, the `msg`/`mcp`/`rest` wiring, tests, and if
-  you add config wiring in `Bridged.java`). Nothing else.
+  you add config wiring in `Fleetd.java`). Nothing else.
 - Work only inside your assigned worktree on your feature branch. The primary fast-forwards `main` after
   re-gating — do not touch `main`.
 - Java 25 idioms are welcome (unnamed `_` params, records). Keep the diff minimal and match surrounding style.
@@ -171,7 +171,7 @@ removal, msgId dedup, cross-restart redelivery). It is `@Tag("contract")`, so th
 untouched. Run it explicitly when Docker (or a broker) is available:
 
 ```bash
-cd bridged
+cd fleetd
 mvn -Pcontract test -Dtest=AmqpReplyInboxContractTest   # local: spins a RabbitMQ Testcontainers fixture
 ```
 

docs/CB-308-Multi-Host-Federation.md

@@ -18,7 +18,7 @@ The design rests on three pieces (the shape this ticket proposes):
 1. **Dedicated per-agent channels** — every agent has its own addressable inbox on the broker.
 2. **A federated agent directory** — a global "who/where/status" lookup, assembled from per-host
    presence, not a central database.
-3. **A per-host gateway** — each host runs a `bridged` that owns its local herdr, registers/manages
+3. **A per-host gateway** — each host runs a `fleetd` that owns its local herdr, registers/manages
    its own sessions, and proxies messages to/from other hosts over the broker.
 
 ## 2. What is single-host today (the assumptions to break)
@@ -27,7 +27,7 @@ The design rests on three pieces (the shape this ticket proposes):
 flowchart TB
     subgraph host["Single host (today)"]
         primary["primary<br/>(MCP client)"]
-        daemon["bridged daemon<br/>127.0.0.1:8765"]
+        daemon["fleetd daemon<br/>127.0.0.1:8765"]
         reg["in-process registry<br/>keyed by PeerHandle.id() == paneId"]
         herdr["herdr<br/>(local unix-socket PTY mux)"]
         w1["worker pane wQ:p1"]
@@ -48,21 +48,21 @@ Three concrete bake-ins assume one host:
 |---|---|---|
 | **herdr is local** | `herdr/` unix socket `~/.config/herdr/herdr.sock` | You cannot drive another host's PTYs → each host **must** own its herdr. This is why a per-host gateway is mandatory. |
 | **registry is in-process, keyed by `paneId`** | `session/SessionManager` | `paneId` (e.g. `wQ:p2B`) is a herdr-local coordinate — meaningless off-host. Routing needs a host-unique id. |
-| **loopback, no authn** | `rest/BridgedApp` binds `127.0.0.1:8765` | Fine on one host; the moment a second host can talk to a gateway, that link is a trust boundary. |
+| **loopback, no authn** | `rest/FleetApp` binds `127.0.0.1:8765` | Fine on one host; the moment a second host can talk to a gateway, that link is a trust boundary. |
 
 ## 3. Target architecture
 
 ```mermaid
 flowchart TB
     subgraph hostA["HOST A"]
-        gA["gateway = bridged A"]
+        gA["gateway = fleetd A"]
         regA["local registry + herdr"]
         primary["primary (MCP client)"]
         gA --- regA
         primary --- gA
     end
     subgraph hostB["HOST B"]
-        gB["gateway = bridged B"]
+        gB["gateway = fleetd B"]
         regB["local registry + herdr"]
         wb["worker panes"]
         gB --- regB
@@ -96,13 +96,13 @@ host's terminals.*
   delayed-message exchange (the remind/backoff loop for free) — the same reasons CB-307 picked it.
 
 - **Federated agent directory** = a **soft-state, bridge-owned** roster, *not* a broker-stored
-  database. Per the persistence-boundary decision (bridged is soft-state; the broker owns *message*
+  database. Per the persistence-boundary decision (fleetd is soft-state; the broker owns *message*
   durability, not *who/where/status*), each gateway announces its local agents `(globalId, host,
   status, capabilities)` on a `roster.*` presence topic with periodic heartbeats. Every gateway
   builds an eventually-consistent **union view** — literally CB-304's `rosterView`, federated. A
   stale entry expires by missed heartbeat (reuses CB-303's idle/TTL thinking).
 
-- **Per-host gateway** = today's `bridged` daemon, evolved. It already registers/manages sessions
+- **Per-host gateway** = today's `fleetd` daemon, evolved. It already registers/manages sessions
   and controls its local herdr; multi-host adds exactly two responsibilities: (a) a broker client
   that consumes its agents' inboxes and injects into local herdr, and (b) presence announce +
   union-roster assembly. Evolution, not rewrite.
@@ -111,7 +111,7 @@ host's terminals.*
 
 ```mermaid
 flowchart LR
-    send["bridge_send(globalId, msg)"] --> lookup{"directory:<br/>is globalId local?"}
+    send["fleet_send(globalId, msg)"] --> lookup{"directory:<br/>is globalId local?"}
     lookup -->|"yes"| local["inject via local herdr<br/>(today's Injector path)"]
     lookup -->|"no"| pub["publish agent.&lt;id&gt;.inbox<br/>(broker routes to owning gateway)"]
     pub --> consume["owning gateway consumes<br/>→ injects into its local herdr"]
@@ -123,7 +123,7 @@ broker. A sender is oblivious to which branch it took.*
 ## 4. What CB-307 already provides vs. what is net-new
 
 **CB-307 delivers the transport half** and is independently valuable on a single host: the AMQP
-broker fabric, the `bridged → broker` client/adapter, at-least-once + idempotent (dedup-by-id)
+broker fabric, the `fleetd → broker` client/adapter, at-least-once + idempotent (dedup-by-id)
 delivery, DLQ, and delayed-retry (remind). That *is* the "proxy cross-host message" backbone;
 extending the same broker from "worker→primary reliability" to "gateway↔gateway" is incremental.
 
@@ -158,17 +158,17 @@ sequenceDiagram
     participant BR as broker
     participant GA as gateway A
     participant P as primary (host A, MCP client)
-    W->>GB: bridge_reply
+    W->>GB: fleet_reply
     GB->>BR: publish primary-bound (durable, msg id)
     BR->>GA: route to A's primary inbox
     Note over GA: held durably until the primary pulls
-    P->>GA: blocking bridge_send resolves / bridge_poll
+    P->>GA: blocking fleet_send resolves / fleet_poll
     GA-->>P: reply (then ACK to broker)
 ```
 
 *Figure 4 — the broker makes the middle hop lossless, ordered, and idempotent; the **final** hop
 into the primary is still a **pull** (gateway A holds the message until the primary's blocking
-`bridge_send` or `bridge_poll`). Cross-host neither improves nor worsens this — it just spans hosts.
+`fleet_send` or `fleet_poll`). Cross-host neither improves nor worsens this — it just spans hosts.
 This is precisely the gap CB-307 closes on one host and CB-308 stretches across hosts.*
 
 ## 6. Staging & dependencies
@@ -207,8 +207,8 @@ they overlap (notably: the envelope is no longer optional, and dedup is split by
    gateway's host. This extends the single-host invariant — *identity comes from the connection,
    never an argument* — across the broker: cross-host, identity comes from the key. Complements
    (not replaces) per-gateway broker logins over TLS.
-2. **Profiles are owned by the worker's host.** `bridge_spawn(profile, host)` resolves the name in
-   the *target* gateway's `bridged.yaml`. Gateways advertise their profile names in presence
+2. **Profiles are owned by the worker's host.** `fleet_spawn(profile, host)` resolves the name in
+   the *target* gateway's `fleetd.yaml`. Gateways advertise their profile names in presence
    heartbeats, so a leader sees what each host offers before spawning; an unknown name is a clear
    error from the target. Secrets (base URLs, tokens) never leave the host that uses them.
 3. **Repo provisioning — clone from the forge, pinned.** A cross-host spawn names the repo URL and
@@ -259,7 +259,7 @@ they overlap (notably: the envelope is no longer optional, and dedup is split by
    serialize acks fleet-wide. Ordering caveat: a *return* (unroutable) arrives **before** the
    confirm, so "confirmed" ≠ "routed"; the sender checks the returned-set at confirm time.
    `mandatory` is false only for `BROADCAST`, where an empty group is legal silence.
-10. **Queue lifecycle is session lifecycle.** `bridge_stop`/reap deletes the worker's inbox queue
+10. **Queue lifecycle is session lifecycle.** `fleet_stop`/reap deletes the worker's inbox queue
     (its `broadcast.*` bindings die with it — no broadcasts to the dead); `x-expires` collects
     queues orphaned by a crashed gateway (long for main/orchestrator inboxes, short for workers).
     Queue names carry a version suffix (`.v2`): AMQP refuses to redeclare an existing durable
@@ -275,11 +275,11 @@ they overlap (notably: the envelope is no longer optional, and dedup is split by
 - **Gateway discovery:** how gateways find the broker and each other (static config vs. discovery).
 - **Control authorization — THE GATE ON U4.** Signing (§7.1) settles *who sent it*; authorization
   is *who may do what*. **Cross-host spawn must not land before the minimal version exists**: a
-  per-host allowlist in `bridged.yaml` — beside the peer public keys — of gateway ids permitted to
+  per-host allowlist in `fleetd.yaml` — beside the peer public keys — of gateway ids permitted to
   publish control to this host, checked against the verified signature. A few lines of config and
   check; without them, any principal holding broker credentials can start processes on every host
   in the fleet.
-- **Key distribution & rotation:** static config (host → public key in each `bridged.yaml`) is
+- **Key distribution & rotation:** static config (host → public key in each `fleetd.yaml`) is
   fine at the current 2–3 host scale; rotation is manual. A refinement, not a blocker.
 - **Gateway death mid-turn:** the roster reaps it by missed heartbeat, and in-flight primary-bound
   messages survive by broker durability; still open is reconciling *worker* state when the dead

docs/CB-401-Peer-Launcher-SPI.md

@@ -7,7 +7,7 @@ the core learning any one peer's environment.
 ## 1. Why
 
 `claude-bridge` is a **communication bus between heterogeneous AI agents** — its stable surface is
-the protocol (`bridge_spawn / send / poll / reply / ask / list / stop / status`), and that surface
+the protocol (`fleet_spawn / send / poll / reply / ask / list / stop / status`), and that surface
 should stay provider-neutral. Today the daemon can only materialize one kind of peer: an
 off-subscription Claude Code CLI over herdr. Everything specific to *how that peer is set up*
 (`ANTHROPIC_BASE_URL`, the subscription guard, `--mcp-config`/system-prompt flags, `claude-*`
@@ -60,7 +60,7 @@ Where Claude/herdr specifics actually live today:
 | `claude-<profile>-<nonce>-<seq>` naming, `WORKER_NAME` regex, orphan reap (CB-117) | `WorkerService` | **→ adapter** (naming is a herdr-label detail) |
 | `GITEA_TOKEN` / `GITEA_HOST` injection (CB-302 checkpoint) | `WorkerService.spawn` | **→ adapter** + a **capability** (§6) |
 | tab/pane placement, worker space, tab labels | `WorkerService.spawnInTab/spawnAsPane` via herdr `WorkspaceControl` | **→ adapter** (herdr transport detail) |
-| `BridgedConfig.Worker` profile shape (`baseUrl`, `model`, `configDir`, …) | `config` | **mostly adapter-shaped** — see §5 |
+| `FleetConfig.Worker` profile shape (`baseUrl`, `model`, `configDir`, …) | `config` | **mostly adapter-shaped** — see §5 |
 | FSM, registry, roster, `reapIdle`/`drainAll`/`contextCap`, `rosterView` | `SessionManager` | **stays core** |
 | turn/completion detection (`TurnListener`, `CompletionResolver`, `StatusPoller`, `WorkerPresence`) | `inject/` | **stays core**, but reads herdr terminal output → transport-coupled (§4b) |
 | message store & routing | `msg/` | **stays core** |
@@ -118,7 +118,7 @@ hard-wire "turns come from herdr".
 
 ## 5. Config shape
 
-`BridgedConfig.Worker` is Claude-shaped (`baseUrl`, `model`, `configDir`, `tokenEnv`). Rather than
+`FleetConfig.Worker` is Claude-shaped (`baseUrl`, `model`, `configDir`, `tokenEnv`). Rather than
 break existing YAML, CB-401 keeps `workers:` exactly as-is and treats those fields as the
 **ClaudeCodeLauncher's** profile schema. A future peer kind adds a `kind:` discriminator
 (default `"claude-code"`) selecting the launcher; unknown-kind → clear config error. No migration of
@@ -126,7 +126,7 @@ existing configs. (Jackson already ignores unknown keys, so adding `kind` is bac
 
 ```mermaid
 sequenceDiagram
-    participant MCP as bridge_spawn (MCP/REST)
+    participant MCP as fleet_spawn (MCP/REST)
     participant SM as SessionManager
     participant L as PeerLauncher (by profile.kind)
     participant T as transport (herdr)
@@ -150,7 +150,7 @@ degrades gracefully when a launcher lacks one:
 
 | Capability | Meaning | Claude Code | Codex (likely) | Human |
 |---|---|---|---|---|
-| `MID_TURN_ASK` | supports `bridge_ask` rendezvous | ✓ | ? | ✗ |
+| `MID_TURN_ASK` | supports `fleet_ask` rendezvous | ✓ | ? | ✗ |
 | `SELF_PR` | can open its own PR at checkpoint (CB-302) | ✓ (opt-in token) | ? | ✗ |
 | `WORKTREE` | can run in a provisioned git worktree | ✓ | ✓ | ✗ |
 | `ORPHAN_REAP` | spawner can reconcile orphaned peers on boot | ✓ | ? | ✗ |
@@ -185,13 +185,13 @@ messages, not verified facts.
 Deliverable for CB-401 Stage A — mechanical, behaviour-preserving:
 
 1. `PeerLauncher` interface + `PeerHandle` (opaque id) + `SpawnRequest` (profile, requestedCwd,
-   callerCwd) + `Capability` enum, new package `dev.ltms.bridged.peer`.
+   callerCwd) + `Capability` enum, new package `dev.ltms.fleet.peer`.
 2. `ClaudeCodeLauncher implements PeerLauncher` = today's `WorkerService`, adapted: `spawn(...)`
    returns a `PeerHandle` (id = paneId), `capabilities()` declares
    `MID_TURN_ASK, SELF_PR(when token), WORKTREE, ORPHAN_REAP`.
 3. `SessionManager` depends on `PeerLauncher`, not `WorkerService` concretely; routing keys on
    `PeerHandle.id()` (== paneId today, so zero value change).
-4. `Bridged.main` wires the concrete `ClaudeCodeLauncher` behind the interface.
+4. `Fleetd.main` wires the concrete `ClaudeCodeLauncher` behind the interface.
 5. **No behaviour change, no config change.** Full green gate: `ide_sync` → `ide_diagnostics`
    (0 errors/0 warnings) → `mvn clean install` with `MVN_EXIT` captured (no masking pipe). All
    existing tests pass unchanged; add tests only for the new `PeerHandle` indirection.

docs/CB-402-OpenCode-Adapter.md

@@ -11,7 +11,7 @@ All five increments of §4 are done, including increment 5 (the §5 live checkli
 
 ## 1. Goal
 
-Prove the [`PeerLauncher`](../bridged/src/main/java/dev/ltms/bridged/peer/PeerLauncher.java) SPI
+Prove the [`PeerLauncher`](../fleetd/src/main/java/dev/ltms/fleet/peer/PeerLauncher.java) SPI
 actually holds for a **non-Claude** coding agent by shipping a second, first-class in-tree
 adapter: **opencode** (`opencode` 1.1.31, a provider-agnostic terminal coding agent).
 
@@ -56,8 +56,8 @@ flowchart TB
 
 *Figure 1 — the two concerns tangled inside today's single launcher; CB-402 splits them.*
 
-There is also a **Stage-A deferral** to finish: `Bridged.main` still casts
-`(ClaudeCodeLauncher) workers` at the `BridgeMcp` and `BridgedApp` constructors. Those two
+There is also a **Stage-A deferral** to finish: `Fleetd.main` still casts
+`(ClaudeCodeLauncher) workers` at the `FleetMcp` and `FleetApp` constructors. Those two
 callers only invoke `profiles()`, `defaultProfile()`, and `list()` — **all already on the
 `PeerLauncher` interface**. The cast survives for one reason only: `PeerLauncher.list()`
 returns `List<?>` (element type erased) while the callers use `Agent` element methods in their
@@ -68,7 +68,7 @@ roster join. Finishing the migration is therefore small and contained (§4.D).
 ## 3. Target design
 
 Template-Method base + two thin adapters + a routing composite that keeps the Stage-A seam
-(one `PeerLauncher` reference held by `SessionManager` / `BridgeMcp` / `BridgedApp`) intact.
+(one `PeerLauncher` reference held by `SessionManager` / `FleetMcp` / `FleetApp`) intact.
 
 ```mermaid
 flowchart TB
@@ -114,7 +114,7 @@ Two adapter **hooks** (abstract):
 protected abstract String namePrefix();               // "claude" | "opencode"
 
 /** Build the peer-specific launch: env map + argv. Runs any pre-spawn guard here. */
-protected abstract Launch buildLaunch(BridgedConfig.Worker cfg, SpawnRequest req);
+protected abstract Launch buildLaunch(FleetConfig.Worker cfg, SpawnRequest req);
 record Launch(Map<String,String> env, List<String> argv) {}
 ```
 
@@ -126,7 +126,7 @@ so the opencode adapter never reaps a `claude-*` pane and vice-versa. The compos
 
 ### B. `kind:` config discriminator
 
-Add one field to `BridgedConfig.Worker`:
+Add one field to `FleetConfig.Worker`:
 
 ```java
 String kind          // "claude-code" (default) | "opencode"
@@ -138,7 +138,7 @@ String kind          // "claude-code" (default) | "opencode"
   so the record stays declarative.)
 - Keep the existing back-compat constructors; `kind` is additive and optional.
 
-`bridged.example.yaml` documents a two-kind `workers:` block.
+`fleetd.example.yaml` documents a two-kind `workers:` block.
 
 ### C. `OpenCodeLauncher` — the adapter hooks for opencode
 
@@ -164,13 +164,13 @@ String kind          // "claude-code" (default) | "opencode"
 
 ### D. `CompositePeerLauncher` + finish the Stage-A migration
 
-- `Bridged.main` groups configured profiles by `kind`, instantiates one launcher per kind
+- `Fleetd.main` groups configured profiles by `kind`, instantiates one launcher per kind
   present, and wraps them in `CompositePeerLauncher implements PeerLauncher`.
 - Routing methods (`spawn(req)`, `effectiveCwd(req)`, `parityOverlay(name)`) dispatch by the
   profile's kind. Fan-out methods (`list()`, `reapOrphanWorkers()`, `capabilities()`,
   `profiles()`, `defaultProfile()`) merge across sub-launchers. `stop(id)` tries each (teardown
   only knows the pane id) — already best-effort/idempotent.
-- **Migrate `BridgeMcp` + `BridgedApp` to the `PeerLauncher` interface**, dropping both
+- **Migrate `FleetMcp` + `FleetApp` to the `PeerLauncher` interface**, dropping both
   `(ClaudeCodeLauncher)` casts. Only friction is `list()`'s `List<?>`; resolve by giving the SPI
   a typed roster element (small neutral `PeerAgent` view exposing `id()`/`name()`/status) that
   the CB-304 roster join consumes — or, minimally, narrow at the callsite. Prefer the typed view.
@@ -179,12 +179,12 @@ String kind          // "claude-code" (default) | "opencode"
 sequenceDiagram
     autonumber
     participant P as Primary
-    participant M as BridgeMcp / REST
+    participant M as FleetMcp / REST
     participant C as CompositePeerLauncher
     participant O as OpenCodeLauncher
     participant B as HerdrPeerLauncher (base)
     participant H as herdr
-    P->>M: bridge_spawn(profile="oc-impl")
+    P->>M: fleet_spawn(profile="oc-impl")
     M->>C: spawn(SpawnRequest)
     C->>C: kind(profile)=="opencode"
     C->>O: spawn(req)
@@ -208,7 +208,7 @@ sequenceDiagram
    `ClaudeCodeLauncher` extend it with `namePrefix()="claude"` and `buildLaunch()` wrapping
    today's guard+env+argv logic. Green build, identical tests — pure refactor. *(IDE
    `refactor` where possible; the primary re-runs the gate workers can't.)*
-2. **`kind:` discriminator.** Add the field + normalization + `bridged.example.yaml`. Default
+2. **`kind:` discriminator.** Add the field + normalization + `fleetd.example.yaml`. Default
    path unchanged (`kind=claude-code`).
 3. **`OpenCodeLauncher`.** Implement the three hooks; unit-test `buildLaunch` (env has no
    `ANTHROPIC_BASE_URL`; `OPENCODE_CONFIG` points at a file carrying the bridge MCP block +
@@ -226,11 +226,11 @@ until it is "major" (Stage-B whole), per the CB-401 bar.
 - **opencode TUI ⇄ herdr injection.** herdr drives a pane by typing into a TUI. Must confirm
   opencode's TUI accepts injected keystrokes/submit the way `claude` does, and reaches an
   `injectable` status the CB-306 gate recognizes. *Validation:* spawn one opencode worker,
-  watch the readiness gate pass, `bridge_send` a trivial task.
+  watch the readiness gate pass, `fleet_send` a trivial task.
 - **Bridge MCP visibility in opencode.** Confirm `OPENCODE_CONFIG` (or `opencode mcp add`)
-  actually surfaces the `bridge_*` tools inside the opencode session, and that `bridge_reply`
+   actually surfaces the `fleet_*` tools inside the opencode session, and that `fleet_reply`
   is callable — the reply-charter is worthless if the tool isn't mounted. *Validation:* the
-  worker completes a task by calling `bridge_reply`; the reply lands via the CB-307 path.
+  worker completes a task by calling `fleet_reply`; the reply lands via the CB-307 path.
 - **opencode MCP/config schema drift.** opencode is fast-moving (1.1.31 today). Pin the config
   schema we generate against the installed version; treat the exact keys (`type: "remote"` vs
   `"http"`, `instructions` shape) as a dogfood-verified fact, not an assumption.
@@ -244,7 +244,7 @@ until it is "major" (Stage-B whole), per the CB-401 bar.
 
 - **Unit (hermetic):** base-extraction regression (existing `ClaudeCodeLauncher` tests pass
   unchanged); `OpenCodeLauncher.buildLaunch` env/argv/config assertions; `kind` normalization
-  in `BridgedConfigTest`; `CompositePeerLauncher` routing + fan-out (merge of `profiles()`,
+  in `FleetConfigTest`; `CompositePeerLauncher` routing + fan-out (merge of `profiles()`,
   summed `reapOrphanWorkers()`, per-kind reap isolation) with fake sub-launchers.
 - **Live (dogfood, manual):** the §5 checklist on the running daemon.
 - **Gate (primary):** IDE diagnostics 0/0 on every changed file, `mvn clean install` green with
@@ -269,7 +269,7 @@ until it is "major" (Stage-B whole), per the CB-401 bar.
 
 ## 8. As-built — live dogfood (2026-07-29)
 
-Run against `bridged` on `127.0.0.1:8766` at main `19cdf8d`, with opencode **1.18.5** installed
+Run against `fleetd` on `127.0.0.1:8766` at main `19cdf8d`, with opencode **1.18.5** installed
 via Homebrew. Every §5 risk is now a verified fact rather than an assumption.
 
 **The version-drift risk was the real one, and it did not bite.** This adapter was designed against
@@ -281,7 +281,7 @@ as a dogfood-verified fact for 1.18.5.
 | §5 risk | Result |
 |---|---|
 | opencode TUI ⇄ herdr injection; CB-306 gate | ✅ `peer pane=wD:p3 reached injectable state` ~0.6s after `agent.start` |
-| Bridge MCP visible + `bridge_reply` callable | ✅ MCP `initialize` from `Implementation[name=opencode, version=1.18.5]`; worker replied through the tool |
+| Bridge MCP visible + `fleet_reply` callable | ✅ MCP `initialize` from `Implementation[name=opencode, version=1.18.5]`; worker replied through the tool |
 | Config schema drift (1.1.31 → 1.18.5) | ✅ unchanged, see above |
 | Provider credentials | ✅ free tier, zero credentials |
 
@@ -291,7 +291,7 @@ Full lifecycle exercised through the REST surface:
    to `OpenCodeLauncher` (`spawning opencode profile=opencode-free`), pane `wD:p3`.
 2. Readiness: `{"ready":true,"status":"idle"}`, roster state `ready`.
 3. `POST /sessions/{id}/message` → **`{"replySource":"reply","reply":"391"}`** — a *structured*
-   `bridge_reply`, not the CB-115 completion-fallback transcript scrape. The clean path.
+   `fleet_reply`, not the CB-115 completion-fallback transcript scrape. The clean path.
 4. `DELETE /workers/wD:p3` → `204`, roster empty, tolerant teardown (`tab_not_found` ignored —
    opencode had already closed its own tab).
 
@@ -301,5 +301,5 @@ identity (loopback peer PID → herdr pane) classified an **opencode** process a
 opencode-specific handling — confirming the identity model is peer-kind-agnostic, which is exactly
 what CB-308 needs when it stretches the roster across hosts.
 
-CB-502 counters for the same run: `bridged_sends_total{outcome="replied"} 1`,
-`bridged_replies_total{path="rendezvous"} 1`, `bridged_inbox_depth{...} 0`.
+CB-502 counters for the same run: `fleet_sends_total{outcome="replied"} 1`,
+`fleet_replies_total{path="rendezvous"} 1`, `fleet_inbox_depth{...} 0`.

docs/CB-500-Multi-Tier-Coordination.md

@@ -38,14 +38,14 @@ never toolchain ownership (§7).
 flowchart TB
     human["human (types)"]
     primary["PRIMARY (Opus)<br/>MCP client — pull-only"]
-    daemon["bridged daemon<br/>127.0.0.1:8765 (single host)"]
+    daemon["fleetd daemon<br/>127.0.0.1:8765 (single host)"]
     comp["CompositePeerLauncher<br/>routes by kind"]
     cc["ClaudeCodeLauncher"]
     oc["OpenCodeLauncher"]
     w1["worker pane (gx00 vLLM)"]
     w2["worker pane (ollama)"]
     human --> primary
-    primary -->|"bridge_send / spawn / ask"| daemon
+    primary -->|"fleet_send / spawn / ask"| daemon
     daemon --> comp
     comp --> cc
     comp --> oc
@@ -80,7 +80,7 @@ flowchart TB
         m1["main A: Opus<br/>MCP client"]
         m2["main B: cloud module<br/>MCP client"]
     end
-    subgraph bus["bridged fabric (CB-307/308 substrate)"]
+    subgraph bus["fleetd fabric (CB-307/308 substrate)"]
         chan["per-agent inbox channels<br/>agent.&lt;globalId&gt;.inbox"]
         roster["federated roster (union view)"]
     end
@@ -146,11 +146,11 @@ env-manager" rule (§7).*
 ```mermaid
 sequenceDiagram
     participant M as main (delegator)
-    participant D as bridged
+    participant D as fleetd
     participant SL as SandboxLauncher
     participant SB as sandbox (peer-owned)
     participant A as agent in sandbox
-    M->>D: bridge_spawn(profile=backend, role=backend)
+    M->>D: fleet_spawn(profile=backend, role=backend)
     D->>SL: spawn(SpawnRequest)
     SL->>SB: start entrypoint (image = backend role)
     Note over SL,SB: bridge injects guarded ANTHROPIC_BASE_URL,<br/>mounts bridge MCP url + reply charter
@@ -189,7 +189,7 @@ flowchart TB
         m2["main B: cloud module<br/>MCP client (pull-only)"]
     end
     reg["PrimaryRegistry → multi-slot<br/>(terminal per main)"]
-    subgraph fabric["bridged"]
+    subgraph fabric["fleetd"]
         ca["agent.A.inbox"]
         cb["agent.B.inbox"]
         push["ReplyPushLoop → N terminals"]
@@ -211,14 +211,14 @@ transport is already peer-neutral — what was missing is N pull endpoints).*
 ```mermaid
 sequenceDiagram
     participant MA as main A (Opus)
-    participant BR as bridged / broker
+    participant BR as fleetd / broker
     participant MB as main B (cloud)
-    MA->>BR: bridge_send(to = main B, msg)
+    MA->>BR: fleet_send(to = main B, msg)
     BR->>BR: publish agent.B.inbox (durable, msg id)
     Note over BR: held until B pulls (B is a client too)
-    MB->>BR: blocking bridge_send / poll resolves
+    MB->>BR: blocking fleet_send / poll resolves
     BR-->>MB: msg (then ACK)
-    MB->>BR: bridge_reply(to = main A)
+    MB->>BR: fleet_reply(to = main A)
     BR->>BR: publish agent.A.inbox
     MA->>BR: poll resolves
     BR-->>MA: reply
@@ -233,7 +233,7 @@ push-loop fan-out; relax "orchestration tools only the primary calls" to "any re
 ## 6. Development C — Orchestrator tier
 
 > **SUPERSEDED — do not implement this model.** The operator rejected a supervisor above the lead.
-> The human continues to drive the pre-existing lead directly; bridged neither spawns nor resumes that
+> The human continues to drive the pre-existing lead directly; fleetd neither spawns nor resumes that
 > lead. What replaced this proposal is **one lead, two short-lived advisory architects, and N workers**:
 > the lead engages architects sideways for a strong-model assessment, then discards them. Architect
 > slots are declared in `architects:` (see Gitea issue #16), rather than making leads managed sessions.
@@ -243,7 +243,7 @@ push-loop fan-out; relax "orchestration tools only the primary calls" to "any re
 
 > **Historical alternative retained.** The text and figures below record the considered model and why it
 > was rejected: it re-rooted the human-facing session above the lead, violating the still-true premise
-> that configured leaders pre-exist, are recognised, and cannot be resumed by bridged.
+> that configured leaders pre-exist, are recognised, and cannot be resumed by fleetd.
 
 The orchestrator is **`SessionManager` recursed one tier up**: today it spawns/names/reaps *worker*
 sessions; the orchestrator does the same for *main* sessions, and adds **context scoping**.
@@ -287,7 +287,7 @@ sequenceDiagram
     H->>O: high-level goal (large context)
     O->>O: name/resume main A session
     O->>MA: task + SCOPED context slice (not the whole history)
-    MA->>W: bridge_send(delegation, carrying only the relevant slice)
+    MA->>W: fleet_send(delegation, carrying only the relevant slice)
     W-->>MA: result
     MA-->>O: rollup
     O->>O: fold into orchestrator context, pick next main/turn
@@ -385,7 +385,7 @@ ticket is an extension of an existing pattern (CB-402 for A, CB-308 for B/C).
 
 The follow-up question — *"clarify the architecture when we have distributed agents in sandboxes"* —
 resolves the fork left open in §4 and §9. **Decision: Development A (sandbox launcher) and CB-308
-(per-host federation) *compose*, not compete — each host runs a `bridged` gateway whose launcher
+(per-host federation) *compose*, not compete — each host runs a `fleetd` gateway whose launcher
 spawns agents into that host's *local* sandboxes.** A sandbox is never reached across the network; it
 is reached by the gateway sitting next to it.
 
@@ -393,7 +393,7 @@ is reached by the gateway sitting next to it.
 
 The bus delivers a turn by **herdr keystroke-injection** — `Injector → AgentControl.send` writes into
 a PTY that its **local** herdr owns. The broker moves *messages and presence*, **never keystrokes**.
-So an agent's PTY must live in a herdr that *some* `bridged` instance drives locally: a remote
+So an agent's PTY must live in a herdr that *some* `fleetd` instance drives locally: a remote
 container with no local herdr **cannot be injected into**. That rules out a central daemon reaching
 remote PTYs, and collapses the design to a single identity:
 
@@ -403,7 +403,7 @@ remote PTYs, and collapses the design to a single identity:
 flowchart TB
     subgraph hostA["HOST A — gateway"]
         mA["main / orchestrator<br/>MCP client → LOCAL gateway"]
-        gA["bridged A<br/>herdr + CompositePeerLauncher<br/>(incl. SandboxLauncher)"]
+        gA["fleetd A<br/>herdr + CompositePeerLauncher<br/>(incl. SandboxLauncher)"]
         cBEa["sandbox: backend<br/>(local container)"]
         cFEa["sandbox: frontend<br/>(local container)"]
         mA --- gA
@@ -415,7 +415,7 @@ flowchart TB
         roster["roster.* (federated presence)"]
     end
     subgraph hostB["HOST B — gateway"]
-        gB["bridged B<br/>herdr + SandboxLauncher"]
+        gB["fleetd B<br/>herdr + SandboxLauncher"]
         cBEb["sandbox: backend<br/>(local container)"]
         gB -->|"spawn → PTY in B's herdr"| cBEb
     end
@@ -440,13 +440,13 @@ sequenceDiagram
     participant BR as broker
     participant GB as gateway B
     participant SB as sandbox agent (host B, container)
-    MA->>GA: bridge_send(globalId on B, msg)
+    MA->>GA: fleet_send(globalId on B, msg)
     GA->>GA: directory lookup - is globalId local? NO
     GA->>BR: publish agent.ID.inbox (durable)
     BR->>GB: route to the owning gateway
     GB->>SB: inject via B's LOCAL herdr (keystrokes)
     Note over GB,SB: SandboxLauncher already spawned the container -<br/>its PTY is in B's herdr, CB-306 readiness passed
-    SB-->>GB: bridge_reply (to B's LOCAL MCP endpoint)
+    SB-->>GB: fleet_reply (to B's LOCAL MCP endpoint)
     GB->>BR: publish primary-bound (durable, msg id)
     BR->>GA: route back to A
     Note over GA: held until the main pulls (the main is a client)
@@ -470,7 +470,7 @@ unchanged — the sandbox is transparent to it.*
 
 | Concern | Provided by |
 |---|---|
-| Per-host gateway (owns local herdr + sessions) | **CB-308** (today's `bridged`, evolved) |
+| Per-host gateway (owns local herdr + sessions) | **CB-308** (today's `fleetd`, evolved) |
 | Spawn into a local sandbox / role→image | **Development A** `SandboxLauncher` (§4), routed by `CompositePeerLauncher` |
 | Addressing a remote sandboxed agent | **CB-308** global id + federated roster (host + role as metadata) |
 | Orphan reap after a gateway restart | **CB-117** per-gateway, summed by the composite — each reaps only its **local** herdr |

docs/CB-591-Gateway-Migration.md

@@ -9,7 +9,7 @@ with no terminator, and our third-party members cannot detect it (§7.2).
 · **Upstream issue:** [systems/vms#31](https://git.ltms.dev/systems/vms/issues/31)
 
 The gateway went live on 2026-08-15 and replaced Bifrost. This plan says what that means for a
-**member definition** in `bridged.yaml`, because that is the part of this repo the change actually
+**member definition** in `fleetd.yaml`, because that is the part of this repo the change actually
 touches.
 
 ---
@@ -51,7 +51,7 @@ Three facts about our side that decide the shape of this work:
    `ANTHROPIC_AUTH_TOKEN` (the value is read from a host env var and never stored in config).
    `local` sets no `tokenEnv` today, because a direct vLLM needs no token.
 2. **`SubscriptionGuard` refuses any host not on an allowlist**, and that allowlist is
-   `guard.offSubscriptionHosts: [gx00.gw]`. It is built once in `Bridged.java:93` and handed to the
+   `guard.offSubscriptionHosts: [gx00.gw]`. It is built once in `Fleetd.java:93` and handed to the
    launcher, so **it is a restart-required key**, not a hot one. Changing `baseUrl` without changing
    this makes every `local` spawn throw.
 3. **The wiki names us as a blocker.** Under *Not done yet*: retiring the shared `legacy` token is
@@ -132,18 +132,18 @@ own provider credentials. So 3b needs **no allowlist change**; only 3a does.
 
 **Both still need a restart, for a different reason.** `tokenEnv` is resolved by
 `HerdrPeerLauncher.resolveEnv` → `env.apply(name)`, which reads the **daemon's own process
-environment**. The running `bridged` inherited its environment when it started, so a variable added to
+environment**. The running `fleetd` inherited its environment when it started, so a variable added to
 `secrets.sh` afterwards is simply not there — the launcher would inject an empty token and the
-gateway would answer 401. This is the same failure as trap 1 in `scripts/redeploy-bridged.sh`
+gateway would answer 401. This is the same failure as trap 1 in `scripts/redeploy-fleetd.sh`
 (`WORKER_GITEA_TOKEN`), and it has the same fix: **restart from a login shell**, and use
-`scripts/redeploy-bridged.sh --check` to confirm the name resolves before restarting anything.
+`scripts/redeploy-fleetd.sh --check` to confirm the name resolves before restarting anything.
 
 ### 3c. What this does to `ccs`
 
 Once a profile carries `baseUrl`, `tokenEnv` and `model` itself, the ccs instance stops being what
 routes a member. Be precise about what is left, though: `configDir` still supplies **folder trust**
 and `settings.json`, and dropping it is what produced the trust dialog and the wrong-model error
-recorded in `bridged.yaml`. So ccs goes from *deciding where the tokens go* to *holding client-side
+recorded in `fleetd.yaml`. So ccs goes from *deciding where the tokens go* to *holding client-side
 state*. Less load-bearing, not removable.
 
 ### Why `/anthropic` and never `/v1/chat/completions`
@@ -186,7 +186,7 @@ Switching buys four things we do not have:
 - **Free opencode capacity, off the shared credential.** The largest single win. See §3b — it retires
   a real single point of failure, not just a cost line.
 - **Per-consumer usage figures.** The cockpit counts requests per consumer. That is the first real
-  measurement of what the fleet consumes, and it feeds [CB-589](https://git.ltms.dev/lms/claude-bridge/issues/74) Gap 2 directly.
+  measurement of what the fleet consumes, and it feeds [CB-589](https://git.ltms.dev/fleet/fleetd/issues/74) Gap 2 directly.
 - **Our own revocable token.** One consumer to revoke if a worker ever leaks it, instead of a shared
   `legacy` token used by four systems.
 - **It works off-LAN.** `gx00.gw` resolves on the LAN only.
@@ -196,7 +196,7 @@ every member spawn. The wiki keeps the direct route open precisely because "if t
 nothing that matters is blocked."
 
 So keep it. Add a second profile `local-direct` pointing at `http://gx00.gw:8000` with **`weight: 0`**
-— never auto-selected, still spawnable with an explicit `bridge_spawn{profile: "local-direct"}`.
+— never auto-selected, still spawnable with an explicit `fleet_spawn{profile: "local-direct"}`.
 That is exactly what CB-554 made `weight: 0` mean, and it turns the escape hatch into something the
 lead can actually reach during an incident.
 
@@ -207,7 +207,7 @@ worker's honesty rule leans on it ("never claim the result of a check you had no
 
 - **Keep bridge-only.** The invariant stays true and simple. Workers stay cheap and narrow.
 - **Add the gateway MCP.** Implementers get context7 documentation lookups, which is genuinely useful
-  for library work. But `mcpUrl` in `BridgedConfig.Profile` is a **single `String`**, so a
+  for library work. But `mcpUrl` in `FleetConfig.Profile` is a **single `String`**, so a
   claude-code member can mount exactly one MCP — this needs a code change, not a config edit.
 
 Note the invariant is **already inaccurate**: `opencode.json` gives `sol` and `terra` both context7
@@ -217,7 +217,7 @@ defensible; picking one is not mine to do.
 ### D3 — token scope
 
 One consumer, `claude-bridge`, its token in `${SHARED_ENV}/tools/secrets.sh` as `AI_GATEWAY_TOKEN`,
-referenced by name only. Never the literal value in `bridged.yaml` — `tokenEnv` exists for this.
+referenced by name only. Never the literal value in `fleetd.yaml` — `tokenEnv` exists for this.
 
 ---
 
@@ -226,7 +226,7 @@ referenced by name only. Never the literal value in `bridged.yaml` — `tokenEnv
 ```mermaid
 flowchart TB
     U1["U1 · consumer token<br/>issue via cockpit, add to secrets.sh"]
-    U2["U2 · profile + guard<br/>bridged.yaml, restart"]
+    U2["U2 · profile + guard<br/>fleetd.yaml, restart"]
     U3["U3 · verify live<br/>spawn, prove thinking survives"]
     U4["U4 · context7 via gateway<br/>.mcp.json + opencode.json"]
     U5["U5 · docs<br/>CLAUDE.md, wiki 11-Features"]
@@ -238,7 +238,7 @@ flowchart TB
 | # | Scope | Who | Why |
 |---|---|---|---|
 | U1 | Issue the `claude-bridge` consumer at `auth.ltms.dev`; store as `AI_GATEWAY_TOKEN` | **operator** | touches secrets and a host we do not own |
-| U2a | New `gx` opencode profile on `/v1` — pure config, no guard change | **lead** | `bridged.yaml` is gitignored, so a worker cannot see or edit it |
+| U2a | New `gx` opencode profile on `/v1` — pure config, no guard change | **lead** | `fleetd.yaml` is gitignored, so a worker cannot see or edit it |
 | U2b | `local` → `/anthropic`; add `local-direct` weight 0; add `llm.ltms.dev` to the guard allowlist | **lead** | same |
 | U2c | One restart from a **login shell**, after U2a and U2b | **lead** | picks up `AI_GATEWAY_TOKEN` into the daemon env *and* the guard allowlist, in one stop |
 | U3 | Live spawn on both new profiles; confirm reasoning survives on each surface | **lead** | needs real spawns and the running daemon |
@@ -266,7 +266,7 @@ Each of these cost someone real debugging time upstream. They apply to us.
 1. **Rotating a token restarts the auth proxy, which drops in-flight streaming responses.** For us
    that means rotating `AI_GATEWAY_TOKEN` kills every live member mid-turn, and an async ticket's
    report goes with it. This is the same rule as a daemon redeploy: **drain the fleet first**
-   (`bridge_list` → `bridge_poll` anything wanted → `bridge_stop`), then rotate.
+   (`fleet_list` → `fleet_poll` anything wanted → `fleet_stop`), then rotate.
 2. **The gateway's own `SecurityPolicy` fails open.** Standalone `aigw run` accepts it and silently
    ignores it — an unauthenticated request returned **200**. Auth is the Caddy proxy in front, and
    nothing else. Never reason as if the gateway authenticates.
@@ -283,10 +283,10 @@ Each of these cost someone real debugging time upstream. They apply to us.
 
 Merging config is not proving it. The checks, in order:
 
-1. `bridge_spawn{profile: "gx"}` succeeds and the member completes a real turn ending in
-   `bridge_reply`. This is the first proof of the token, the URL and the model name, and it risks
+1. `fleet_spawn{profile: "gx"}` succeeds and the member completes a real turn ending in
+   `fleet_reply`. This is the first proof of the token, the URL and the model name, and it risks
    nothing the fleet depends on.
-2. `bridge_spawn{profile: "local"}` succeeds. If the guard allowlist was missed, this **throws** — a
+2. `fleet_spawn{profile: "local"}` succeeds. If the guard allowlist was missed, this **throws** — a
    loud, self-correcting failure, which is the good kind. If the restart was missed, it also throws,
    for the same reason.
 3. A `local` member completes a turn. That exercises streaming through two TLS edges, the auth proxy
@@ -297,9 +297,9 @@ Merging config is not proving it. The checks, in order:
    deltas. For `gx` on `/v1`, this answers the open question in §3 rather than assuming it.
 5. The cockpit at `auth.ltms.dev` shows requests counted against the `claude-bridge` consumer, not
    `legacy`. That is the whole point of taking our own token.
-6. `bridge_spawn{profile: "local-direct"}` still works, so the escape hatch is real rather than
+6. `fleet_spawn{profile: "local-direct"}` still works, so the escape hatch is real rather than
    theoretical.
-7. `bridge_list` shows `gx` carrying no `credentialId`, so a `sol`/`terra` exhaustion cannot
+7. `fleet_list` shows `gx` carrying no `credentialId`, so a `sol`/`terra` exhaustion cannot
    quarantine it. This is the single-point-of-failure claim in §3b, checked rather than asserted.
 
 ---
@@ -372,7 +372,7 @@ one turn; this one costs the whole task and is indistinguishable from a slow wor
 
 > **Diagnosing a stuck opencode member.** Do not read its pane. The launcher writes its config to a
 > temp dir and passes it as `OPENCODE_CONFIG` — find it with
-> `ls -dt /var/folders/*/*/T/bridged-opencode-* | head -1`, check the provider block and the key's
+> `ls -dt /var/folders/*/*/T/fleetd-opencode-* | head -1`, check the provider block and the key's
 > length and prefix (never its value), then reproduce with `opencode run --auto -m <provider>/<model>`
 > using the same `OPENCODE_CONFIG`. That is what turned "it hangs" into a one-line error.
 
@@ -383,7 +383,7 @@ one turn; this one costs the whole task and is indistinguishable from a slow wor
 - `/v1/models` returns exactly `["deepseek-v4-flash"]`, so trap 3 is clear.
 - **Reasoning survives both surfaces** — see §3b above.
 - The launcher's generated opencode provider block is correct, carrying a real 48-character `llmk-`
-  key rather than the `bridged-local-noauth` placeholder.
+  key rather than the `fleetd-local-noauth` placeholder.
 - `SubscriptionGuard` accepted `llm.ltms.dev` after the allowlist edit and the restart: `local`
   spawned without throwing, which is the check that catches a missed restart.
 
@@ -460,7 +460,7 @@ code.** That is the whole reason this section exists.
 
 ## 8. Related
 
-- [CB-589 / #74](https://git.ltms.dev/lms/claude-bridge/issues/74) — cost-first placement and a
+- [CB-589 / #74](https://git.ltms.dev/fleet/fleetd/issues/74) — cost-first placement and a
   gateway that reports live capacity. The per-consumer figures this migration unlocks are the first
   input that ticket actually needs.
 - `docs/CB-500-Multi-Tier-Coordination.md` §11 — the distributed-sandbox topology this gateway is

docs/CB-5xx-Hardening.md

@@ -12,12 +12,12 @@ not after it.
 
 ## 1. Why this stage is not optional bookkeeping
 
-`bridged` today has **exactly one security control: the loopback bind**. Every other guarantee
+`fleetd` today has **exactly one security control: the loopback bind**. Every other guarantee
 rests on it.
 
 The identity model (`mcp/ConnectionIdentity.java`) resolves a caller from the connection alone —
 the OS reports the connecting PID, herdr owns the PID→pane map, so a worker cannot forge another
-worker. Its own javadoc is explicit: *"Single-host only (the herd shares the `bridged` host); the
+worker. Its own javadoc is explicit: *"Single-host only (the herd shares the `fleetd` host); the
 token path is the split-host fallback."* The token path does not exist yet.
 
 That leaves a seam that is **latent today and load-bearing the moment the bind moves**:
@@ -67,7 +67,7 @@ it will be disabled and the stage is wasted. So:
 auth:
   mode: loopback-trust   # default — behaves exactly like today: loopback ⇒ PRIMARY, no token needed
   # mode: token          # every non-worker caller must present a valid bearer token
-  # tokenEnv: BRIDGED_API_TOKEN   # host env var holding the token; never the literal value
+  # tokenEnv: FLEETD_API_TOKEN   # host env var holding the token; never the literal value
 ```
 
 `mode: loopback-trust` is the current behaviour, named honestly and now *chosen* rather than
@@ -112,9 +112,9 @@ The roadmap says "systemd unit". **This host is macOS — there is no systemd on
 not found), and the daemon that has been dogfooded for weeks runs as a bare foreground
 `java -jar`. Ship **both**:
 
-- `deploy/dev.ltms.bridged.plist` — launchd agent, the *actual* runtime here, with `KeepAlive` and
+- `deploy/dev.ltms.fleet.plist` — launchd agent, the *actual* runtime here, with `KeepAlive` and
   ordered start after herdr.
-- `deploy/bridged.service` — systemd unit for the Linux gateways CB-308 introduces.
+- `deploy/fleetd.service` — systemd unit for the Linux gateways CB-308 introduces.
 
 Ordering after herdr is advisory in both: the herdr socket may not exist at boot, so the daemon
 must **retry the socket rather than exit** — supervision ordering is a nicety, socket-retry is the
@@ -158,10 +158,10 @@ configured. It already leaks nothing but herdr's version and up/down.
 ### 3.1 There are TWO entry paths, and only one of them has identity today
 
 The wiki describes MCP as "a thin adapter over the REST core". **At the code level that is not
-literally true, and the difference is security-relevant.** `BridgeMcp` calls `MessageService` /
+literally true, and the difference is security-relevant.** `FleetMcp` calls `MessageService` /
 `SessionManager` *directly*; it never issues an HTTP request against a Javalin route. And `/mcp` is
 mounted as a raw servlet on Jetty's `ServletContextHandler`
-(`BridgedApp.build → cfg.jetty.modifyServletContextHandler`), so it does **not** pass through
+(`FleetApp.build → cfg.jetty.modifyServletContextHandler`), so it does **not** pass through
 Javalin's `before` filters at all.
 
 The current split is the mirror image of what you'd expect:
@@ -172,7 +172,7 @@ The current split is the mirror image of what you'd expect:
 | REST routes | ❌ **none at all** — the session id is taken from the URL path and trusted | ❌ none |
 
 So REST is the *more* exposed surface: `POST /sessions/{id}/reply` accepts any `{id}` from the
-path, whereas the MCP `bridge_reply` derives the worker from the connection and refuses to read it
+path, whereas the MCP `fleet_reply` derives the worker from the connection and refuses to read it
 from an argument. Loopback-only bind is what makes this safe today.
 
 **Therefore CB-505 must enforce on both paths against one shared resolver** — not at a single
@@ -188,15 +188,15 @@ Deliberately small; every one maps to a failure mode we have actually hit.
 
 | Metric | Type | Why it exists |
 |---|---|---|
-| `bridged_sends_total{outcome}` | counter | outcome ∈ replied\|completion_fallback\|timeout\|failed — the completion-fallback rate is the health signal for turn detection (CB-115/116/118) |
-| `bridged_send_duration_seconds` | histogram | delegated turn latency |
-| `bridged_replies_total{path}` | counter | path ∈ rendezvous\|inbox — how often a reply strands (CB-307's whole reason to exist) |
-| `bridged_inbox_depth{target}` | gauge | undrained replies; steady-state should be 0 |
-| `bridged_push_nudges_total{outcome}` | counter | outcome ∈ delivered\|exhausted — a rising `exhausted` means the primary is not draining |
-| `bridged_spawns_total{kind,outcome}` | counter | outcome ∈ ready\|timeout\|guard_rejected; per peer kind (CB-402) |
-| `bridged_sessions{state}` | gauge | SPAWNING/READY/BUSY/DONE census |
-| `bridged_herdr_calls_total{method,outcome}` | counter | socket health — the dependency everything rests on |
-| `bridged_auth_failures_total{reason}` | counter | only meaningful once CB-501 lands; catches misconfigured workers |
+| `fleet_sends_total{outcome}` | counter | outcome ∈ replied\|completion_fallback\|timeout\|failed — the completion-fallback rate is the health signal for turn detection (CB-115/116/118) |
+| `fleet_send_duration_seconds` | histogram | delegated turn latency |
+| `fleet_replies_total{path}` | counter | path ∈ rendezvous\|inbox — how often a reply strands (CB-307's whole reason to exist) |
+| `fleet_inbox_depth{target}` | gauge | undrained replies; steady-state should be 0 |
+| `fleet_push_nudges_total{outcome}` | counter | outcome ∈ delivered\|exhausted — a rising `exhausted` means the primary is not draining |
+| `fleet_spawns_total{kind,outcome}` | counter | outcome ∈ ready\|timeout\|guard_rejected; per peer kind (CB-402) |
+| `fleet_sessions{state}` | gauge | SPAWNING/READY/BUSY/DONE census |
+| `fleet_herdr_calls_total{method,outcome}` | counter | socket health — the dependency everything rests on |
+| `fleet_auth_failures_total{reason}` | counter | only meaningful once CB-501 lands; catches misconfigured workers |
 
 ---
 
@@ -226,7 +226,7 @@ was answered from the forge itself. Kept here as the decision record.*
 
 1. ✅ **TLS scope (D3) — confirmed.** Bearer auth + the fail-fast bind guard ship in the daemon;
    TLS terminates at a reverse proxy, documented with a worked example. AMQP gets TLS via an
-   `amqps://` URI. No keystore handling in `bridged`.
+   `amqps://` URI. No keystore handling in `fleetd`.
 2. ✅ **Micrometer (D4) — confirmed dropped.** Zero-dependency Prometheus text renderer, for the
    reasons in D4 (pom reconciliation burden + the mandated CVE gate being un-runnable this
    session). Revisit if a push-gateway or JVM-metrics requirement appears; the endpoint is the

docs/M4-Fleet-Health.md

@@ -1,7 +1,7 @@
 # M4 - Fleet health, recovery, routing, and capacity
 
 **Status:** Design accepted on 2026-08-15. CB-573 part 1 has shipped the classification model and
-the `bridge_list` capacity view; the remaining M4 units are not yet shipped. See
+the `fleet_list` capacity view; the remaining M4 units are not yet shipped. See
 [Unit 2 — what has landed so far](#unit-2---what-has-landed-so-far) before planning Unit 2 work:
 some of its criteria were met by separate CB tickets, and one of them contradicts the unit text.
 **Scope:** Fleet evidence, safe mechanical repair, lead routing, capacity reporting, and optional
@@ -72,7 +72,7 @@ post-teardown invariant so a later regression becomes `DELEGATION_ORPHANED`.
 
 ### 2.2 Corrections made during design
 
-The first state table missed `BUSY` in bridged plus `IDLE` or `DONE` in herdr. It would have found
+The first state table missed `BUSY` in fleetd plus `IDLE` or `DONE` in herdr. It would have found
 the real trace only through a late, weak stall timer. The final model adds
 `TURN_BOUNDARY_LOST` as a strong disagreement state.
 
@@ -145,7 +145,7 @@ Idle may drive configured resource cleanup. It never opens an incident and never
 | `TURN_BOUNDARY_LOST` | Same session turn stays `BUSY`; same accepted task stays open; two raw snapshots show `IDLE` or `DONE` | Strong disagreement. Strict reconciliation may repair it. |
 | `ERROR_ON_SCREEN` | Suspicious non-working state survives grace; `detection` matches a tested adapter-specific fatal signature | Certain only for the matched signature. A bare word such as `Exception` is not enough. |
 | `STALL_SUSPECTED` | Open turn is older than the configured threshold; two normalised `recent_unwrapped` digests are unchanged; no boundary or reply occurs | Not certain. A long valid API call can look the same. Lead decides. |
-| `MUTE` | Turn resolves through completion fallback instead of `bridge_reply` | Certain that no structured reply won. It does not prove an MCP failure. A single event is a metric, not an incident. |
+| `MUTE` | Turn resolves through completion fallback instead of `fleet_reply` | Certain that no structured reply won. It does not prove an MCP failure. A single event is a metric, not an incident. |
 | `REPLY_STRANDED` | Typed reply or health message remains after owning-lead push reaches its cap | Collection failed. This does not explain whether the lead is busy, dead, or ignoring the nudge. |
 | `DELEGATION_ORPHANED` | Target is gone, failed, or released, but one or more tasks remain `PENDING` after reconciliation grace | Certain bridge invariant failure. This is not an inbox-drain fault. |
 | `WORK_PRODUCT_AT_RISK` | Provisioned branch has commits after its recorded base; member is `DONE`, `FAILED`, or preserved after release; no turn or inbox item remains; long-idle threshold passed | A warning, not proof of loss. Work may already have an open pull request or a squash merge. |
@@ -161,7 +161,7 @@ request or merge state. If committed work appears with `REPLY_STRANDED` or
 
 | State | Exact evidence | Certainty and action |
 |---|---|---|
-| `CONTROL_LINK_DOWN` | Two full-fleet `agent.list` calls fail across the grace, and herdr `ping` also fails | Certain for the bridged-to-herdr link. Retry calls, record the incident, and use human escalation if no lead can be reached. |
+| `CONTROL_LINK_DOWN` | Two full-fleet `agent.list` calls fail across the grace, and herdr `ping` also fails | Certain for the fleetd-to-herdr link. Retry calls, record the incident, and use human escalation if no lead can be reached. |
 
 A failed fleet list alone is not a dead-member claim. A single `_not_found` with a healthy global
 link is a target fault, not a control-link fault.
@@ -230,8 +230,8 @@ Only the lead may:
 - choose how to use partial work in a worktree;
 - restart herdr or change network, model, credentials, backend, or configuration.
 
-Reports include literal safe tool calls such as `bridge_status(sessionId="...")`,
-`bridge_poll(ticket="...")`, `bridge_list()`, and optional `bridge_stop(paneId="...")`. A judgement
+Reports include literal safe tool calls such as `fleet_status(sessionId="...")`,
+`fleet_poll(ticket="...")`, `fleet_list()`, and optional `fleet_stop(paneId="...")`. A judgement
 state never presents stop as the only action.
 
 ### 5.3 Release causes and worktree safety
@@ -251,7 +251,7 @@ preserving cause.
 
 Before abnormal release removes the live session, M4 writes an atomic manifest under the worktree
 root. It records session identity, owner, role, profile, repository, path, branch, base commit,
-release cause, release time, state, and pending task ids. `bridge_list.preservedWorktrees` loads these
+release cause, release time, state, and pending task ids. `fleet_list.preservedWorktrees` loads these
 manifests after restart. Stop output and WARN logs also name the path and cause. M4 never
 auto-deletes a preserved worktree.
 
@@ -298,7 +298,7 @@ Compiled brakes apply even if config asks for more:
 - at most two pane reads occur in one fleet tick;
 - targets rotate fairly;
 - only a normalised digest and optional clipped local excerpt are stored;
-- no pane excerpt leaves bridged in a human webhook.
+- no pane excerpt leaves fleetd in a human webhook.
 
 Use `detection` for tested screen signatures. Use normalised `recent_unwrapped` only for progress
 comparison.
@@ -330,7 +330,7 @@ The reader uses AMQP `content_type`, never body sniffing:
 
 ```text
 Legacy v0: text/plain
-Typed family: application/vnd.ltms.bridged.inbox-message+json
+Typed family: application/vnd.ltms.fleet.inbox-message+json
 ```
 
 A legacy reply may begin with `{`. It remains plain text because its media type is `text/plain`.
@@ -348,7 +348,7 @@ Invalid known-format data is copied byte-for-byte to durable queue
 before the original is acknowledged. A failed quarantine handoff leaves the original unacknowledged.
 The raw body never enters logs.
 
-Decode failure creates a redacted WARN, metric, `bridge_list` summary, and routed health incident.
+Decode failure creates a redacted WARN, metric, `fleet_list` summary, and routed health incident.
 One bad entry never escapes the consumer callback and never stops later valid messages.
 
 Safe downgrade is not supported. The previous build ignores `content_type` and would show typed JSON
@@ -399,13 +399,13 @@ Choose the candidate with the fewest assigned foreign incidents. Break ties by s
 then terminal id. Pin the recipient. Reassign only if that peer becomes unhealthy or retires. A
 routing generation marks a reassignment, and old pending assignments become superseded.
 
-`bridge_list` lead rows show health, health age, assigned foreign incident count, and a bounded list
+`fleet_list` lead rows show health, health age, assigned foreign incident count, and a bounded list
 of incident id, subject, state, severity, age, and routing generation. The top-level view also shows
 owner, recipient, and routing reason.
 
 A peer incident is published under the recipient lead's inbox key, not the failed subject's key. Its
 status-gated nudge names the failed lead and gives the exact
-`bridge_poll(target="<recipient-terminal>")` call.
+`fleet_poll(target="<recipient-terminal>")` call.
 
 ### 7.5 Lead inbox ownership
 
@@ -427,7 +427,7 @@ An idle, reachable lead may receive the existing bounded nudge. An unreachable o
 has no safe in-loop recovery. The bridge must not restart or replace it. A new lead would not have the
 failed lead's plan or context, and an uncertain relaunch could create two orchestrators.
 
-With no webhook, only `bridge_list`, `/healthz`, metrics, WARN logs, and the incident journal remain.
+With no webhook, only `fleet_list`, `/healthz`, metrics, WARN logs, and the incident journal remain.
 These are passive surfaces. They are not a human notification.
 
 ## 8. Lost-boundary reconciliation
@@ -520,8 +520,8 @@ A repaired result uses distinct `RECONCILED_COMPLETION` values in `Rendezvous`,
 task poll source, and metrics. The lead sees:
 
 ```text
-[repaired completion - bridged detected a lost turn boundary. The member did not call
-bridge_reply; pane-derived text follows and may be partial]
+[repaired completion - fleetd detected a lost turn boundary. The member did not call
+fleet_reply; pane-derived text follows and may be partial]
 ```
 
 Clipped text also keeps the existing clipped-tail marker.
@@ -551,7 +551,7 @@ failure operation once. It never recreates the task.
 
 Capacity is a view, not a health state.
 
-`bridge_list` adds one block per profile:
+`fleet_list` adds one block per profile:
 
 ```text
 profile, maxLoad, live, free, reclaimable
@@ -612,7 +612,7 @@ incident feeds lead-health evidence. If the lead then becomes unhealthy, peer or
 Webhook mode requires a resolved environment variable. Turning notification off stops outbound
 attempts but keeps incidents. Turning it back on resumes still-open human incidents.
 
-Without a sink, `bridge_list.healthCoverage` states that human escalation is unavailable. `/healthz`
+Without a sink, `fleet_list.healthCoverage` states that human escalation is unavailable. `/healthz`
 keeps its existing HTTP liveness result and adds a nested `fleetHealth.status=partial` component.
 Metrics and one startup or reload WARN expose the same limit.
 
@@ -687,14 +687,14 @@ The sink response body is ignored. A webhook cannot direct recovery. n8n remains
 M4 adds bounded-label series:
 
 ```text
-bridged_health_incidents{scope,state,severity}
-bridged_health_incidents_total{event}
-bridged_health_notifications_total{event,outcome}
-bridged_health_notification_queue_depth
-bridged_health_notification_last_success_seconds
-bridged_health_notification_capability{mode,status}
-bridged_lead_health{lead,state}
-bridged_lead_assigned_incidents{lead}
+fleet_health_incidents{scope,state,severity}
+fleet_health_incidents_total{event}
+fleet_health_notifications_total{event,outcome}
+fleet_health_notification_queue_depth
+fleet_health_notification_last_success_seconds
+fleet_health_notification_capability{mode,status}
+fleet_lead_health{lead,state}
+fleet_lead_assigned_incidents{lead}
 ```
 
 Metric labels never include terminal ids, incident ids, URLs, or error text.
@@ -792,7 +792,7 @@ Acceptance criteria:
     them.
 16. Explicit stop is state-aware. Any pending task or non-terminal state preserves the worktree.
 17. Atomic preserved-worktree manifests reload after restart and appear in lead-only
-    `bridge_list.preservedWorktrees`.
+    `fleet_list.preservedWorktrees`.
 18. Manifest failure preserves the worktree and opens an operator-visible health failure.
 19. Provision records the base commit. Terminal, long-idle worktrees report
     `WORK_PRODUCT_AT_RISK` only under the evidence in Section 4.2 and never auto-delete work.
@@ -807,7 +807,7 @@ Checked against `main` at `e09cac6` on 2026-08-15. Unit 2 was written as one blo
 have since been built by separate CB tickets. Read this before planning the rest, or that work gets
 done twice.
 
-The check was a symbol survey of `bridged/src/main/java` plus the merge history. It tells you whether
+The check was a symbol survey of `fleetd/src/main/java` plus the merge history. It tells you whether
 the machinery exists at all. It is **not** a line-by-line audit of whether each criterion is fully
 met, and I did not run one.
 
@@ -820,7 +820,7 @@ met, and I did not run one.
 | 14 | `DELEGATION_ORPHANED` | 3 files | **Partial.** The health state exists. The teardown-invariant check that creates it, and the retry rule, do not. |
 | 15 | `SPAWN_ROLLBACK` | 0 files | **Contradicted — see below.** |
 | 16 | — | — | Partial at best. CB-576 made release preserve a dirty worktree; whether explicit stop is state-aware is not checked. |
-| 17, 18 | `preservedWorktrees` | 0 files | Not started. No manifest, and no lead-only `bridge_list` field. |
+| 17, 18 | `preservedWorktrees` | 0 files | Not started. No manifest, and no lead-only `fleet_list` field. |
 | 19 | `WORK_PRODUCT_AT_RISK` | 0 files | Not started. |
 
 **Criterion 15 no longer matches the code, and the code is right.** It says "normal `COMPLETED`
@@ -837,7 +837,7 @@ preserve it. `SPAWN_ROLLBACK` itself does not exist yet.
 ### Unit 3 - Typed inbox and member routing
 
 Scope: semantic record, AMQP migration, both adapters, member routing, polling, and member health in
-`bridge_list`.
+`fleet_list`.
 
 Acceptance criteria:
 
@@ -852,7 +852,7 @@ Acceptance criteria:
 6. Invalid known data never escapes the callback, appears as a reply, or blocks later valid messages.
 7. Invalid data reaches durable per-target quarantine before original ack. Failed handoff leaves the
    original unacked.
-8. Decode failures create redacted WARN, metric, `bridge_list` summary, and routed incident without
+8. Decode failures create redacted WARN, metric, `fleet_list` summary, and routed incident without
    raw content.
 9. Both adapters pass one semantic contract for fields, FIFO, dedup, ownership, ack, and release.
 10. Lead keys require explicit ownership. Publication never claims a queue.
@@ -865,7 +865,7 @@ Acceptance criteria:
 15. RabbitMQ contract tests pass with `mvn test -Pcontract`. The same cases run once on production
     LavinMQ, or the release states that LavinMQ was not checked.
 16. Member incidents route to the exact delegating lead and never resolve a task rendezvous.
-17. `bridge_list` shows compact member health and capacity without pane content. Member
+17. `fleet_list` shows compact member health and capacity without pane content. Member
     `idleForSeconds` is present only when no accepted turn or inbox item exists.
 
 ### Unit 4 - Lead health and peer routing
@@ -887,7 +887,7 @@ Acceptance criteria:
 8. A selected working peer is not interrupted. Its push waits for an injectable window.
 9. Recipient assignment stays pinned. Reassignment increments generation and supersedes old pending
    assignment.
-10. `bridge_list` shows bounded foreign assignments, recipient, reason, and generation without pane
+10. `fleet_list` shows bounded foreign assignments, recipient, reason, and generation without pane
     content.
 11. `LeadInboxRegistry` owns configured and discovered lead keys before publication.
 12. Missing leads keep ownership. Retirement needs an empty queue and handled incidents.
@@ -911,7 +911,7 @@ Acceptance criteria:
 3. Webhook mode requires a resolved environment value. Bad notification config does not disable an
    already valid detector.
 4. Mode changes keep open incidents. Re-enable resumes eligible incidents.
-5. `bridge_list`, `/healthz`, metrics, and one WARN show partial coverage without a sink. HTTP
+5. `fleet_list`, `/healthz`, metrics, and one WARN show partial coverage without a sink. HTTP
    liveness behavior stays unchanged.
 6. One-lead, no-sink coverage states that lead failure has no active notification or recovery.
 7. Incident and outbound dedupe use the stable keys in Section 10.3.
@@ -927,7 +927,7 @@ Acceptance criteria:
 15. Webhook response bodies are ignored and cannot direct recovery.
 16. Tests cover disabled mode, one lead without sink, open/update/reminder/resolve, restart, dedup,
     reassignment, disable/re-enable, and sink failure while local health continues.
-17. `bridged.example.yaml` documents all hot keys and compiled floors.
+17. `fleetd.example.yaml` documents all hot keys and compiled floors.
 18. The operator Features wiki is updated separately. The portable `CLAUDE.md` block is checked and
     changed only if shipped tool or inbox semantics make it untrue.
 19. `mvn clean install` passes.

docs/MCP-Contract.md

@@ -1,311 +1,162 @@
-# MCP Contract — `bridged`'s unified gateway
+# MCP flows and error model — `fleetd`
 
-> **Status: 🔴 HISTORICAL DESIGN — do NOT use as the tool reference.** Written 2026-07-14, before
-> any MCP code existed. The system shipped and this page never caught up, so **its tool names,
-> parameter names and REST paths are wrong today**. Audited 2026-08-17; the specific drift:
+> **What this page is.** The **flows**: how a delegation, a clarification, a detached task and a
+> silent member each travel through `fleetd`. These shapes are what shipped, and they are hard to
+> read off the code because they span the MCP face, the rendezvous registry, the `Injector` and
+> herdr.
 >
-> - **Tools it names that do not exist:** `bridge_read`, `bridge_cancel`.
-> - **Shipped tools it omits:** `bridge_poll`, `bridge_ack`, `bridge_profiles`, `bridge_whoami`.
-> - **Parameter names are wrong nearly everywhere** — it says `message`/`target`/`timeout_seconds`/
->   `block` where the code takes `content`/`sessionId`/`timeoutMs`/`wait`; `text` where
->   `bridge_reply` takes `content`; `target` where `bridge_stop` takes `paneId`.
-> - **REST paths are wrong:** it says `POST /workers` and `DELETE /workers/{paneId}`; the daemon
->   serves `POST /members` and `DELETE /members/{paneId}`.
+> **What this page is NOT: a tool reference.** It deliberately holds no tool catalogue, no
+> parameter tables and no REST paths. **The live MCP schema is the authority** — each tool's own
+> description and parameters, as mounted — with the intent→tool table in `CLAUDE.md` as the short
+> form.
 >
-> **The authoritative tool surface is the live MCP schema** (each tool's own description and
-> parameters, as mounted), with the intent→tool table in `CLAUDE.md` as the short form. Both were
-> checked against `mcp/BridgeMcp.java` on 2026-08-17 and are accurate.
+> That absence is the fix for fleetd #114 (CB-609), and it is worth stating why. This page used to
+> carry a full tool catalogue written in July 2026, before any MCP code existed. The code shipped;
+> the page did not follow. By August it named two tools that do not exist, omitted five that do,
+> had the wrong name for nearly every parameter, pointed at REST paths the daemon does not serve,
+> and — worst — still described an identity model (*"any connection that does not map to a known
+> worker is treated as a primary"*) that was a real privilege bug, fixed since by the ancestry
+> walk in fleetd #161. Every one of those errors is the same error: **a second, hand-maintained
+> copy of something the code already states**. So the second copy is gone rather than corrected.
+> Only the flows remain, because a flow is a shape rather than a name, and shapes are what this
+> page was ever good for.
 >
-> What is still worth reading here is **§6 — the flows and the error model** (rendezvous,
-> `bridge_ask`, detached delivery, the turn-done fallback). The shapes it describes are the ones
-> that shipped; only the names around them drifted. Rewriting this page is tracked as **CB-609**.
-
-`bridged` is the **sole communication gateway** for every Claude session in the bridge. Both
-the **primary** (Opus, on subscription) and every **worker** (off-subscription Claude Code)
-mount the *same* MCP server with a single `claude mcp add` line, and talk only through its
-tools. No Claude session ever addresses a broker, a peer, or the network directly.
-
-This document defines every MCP tool that face must expose, who may call it, its blocking
-semantics, and how it maps onto the code already in the tree.
+> The names that do appear below are checked by `McpContractDocTest`, which fails if this page
+> names a `fleet_*` tool the server does not register. That test is the whole reason it is safe to
+> write a tool name here at all.
 
 ---
 
-## 1. Design constraints (non-negotiable)
+## 1. Rendezvous flows
 
-These come from the project's core invariants and bound every decision below.
+### 1.1 Delegation — happy path
 
-1. **One server, both roles.** The primary and all workers mount an identical server. The
-   catalog must serve both, and `bridged` must decide *who is calling* from the connection —
-   never from a caller-supplied argument that could be spoofed.
-2. **Subscription-safe by construction.** No MCP tool ever reads, sets, or forwards
-   `ANTHROPIC_BASE_URL`. Mounting the bridge cannot move a session off subscription.
-   Enforced today by [`SubscriptionGuard`](1-Architecture).
-3. **Blocking rendezvous, no busy-poll.** The primary consumes a worker's reply through a
-   *single* MCP call that `bridged` holds open — never a cross-turn poll loop that would burn
-   subscription quota.
-4. **Status-gated delivery.** Anything that puts text into a worker flows through the existing
-   [`Injector`](1-Architecture): delivered only when the worker is `idle`/`blocked`, at most
-   one message per turn.
-5. **`bridged` owns policy; herdr owns PTYs.** MCP tools express *intent*; `bridged`
-   translates it into guard checks, rendezvous bookkeeping, and herdr `agent.*` calls.
-
----
-
-## 2. Topology
-
-Both faces live in the one daemon. The **north face** is MCP (this document); the **south
-face** is the herdr Unix socket. REST/SSE remains only for non-Claude clients and dashboards.
-
-```mermaid
-flowchart LR
-    OPUS["Opus — primary<br/>(Claude Code, env CLEAN)<br/>MCP client"]
-    subgraph BD["bridged — standalone daemon"]
-        MCP["MCP server (north face)<br/>bridge_send · bridge_reply<br/>bridge_ask · bridge_status · lifecycle"]
-        RDV["rendezvous registry<br/>(blocking-call waiters)"]
-        INJ["Injector + StatusPoller<br/>(status-gated writer)"]
-        SOCK["herdr socket client (south face)"]
-        MCP --> RDV
-        RDV --> INJ
-        INJ --> SOCK
-        MCP --> SOCK
-    end
-    HERDR["herdr<br/>panes · agent-status"]
-    W["worker claude pane<br/>ANTHROPIC_BASE_URL set<br/>MCP client"]
-
-    OPUS -->|"bridge_send (blocks)"| MCP
-    W -.->|"bridge_reply / bridge_ask"| MCP
-    SOCK -->|"agent.start · agent.send<br/>agent.get · pane.close"| HERDR
-    HERDR -->|"drives PTY"| W
-
-    classDef ext fill:#2b6cb0,stroke:#1a365d,color:#ffffff;
-    classDef core fill:#2f855a,stroke:#22543d,color:#ffffff;
-    class OPUS,W ext
-    class MCP,RDV,INJ,SOCK core
-```
-
----
-
-## 3. Identity & addressing
-
-Because the same server is mounted by everyone, `bridged` resolves the caller's role on every
-request — this is the linchpin of the whole contract and has no code yet.
-
-- **Workers are known.** `bridged` spawns every worker
-  ([`WorkerService`](1-Architecture)) and records its herdr session UUID / `terminal_id` on
-  the returned [`Agent`]. When a call arrives on a connection that maps to a known worker,
-  the caller is *that* worker — so **workers never pass a target**; routing is implicit.
-- **The primary is "not a worker".** Any connection that does not map to a known worker is
-  treated as a primary. It addresses workers **explicitly** by `target` — a session UUID,
-  a `terminal_id`, or a friendly `profile` name.
-- **Turn correlation.** A blocking `bridge_send` registers a *waiter* keyed by worker
-  identity. A worker's later `bridge_reply` / `bridge_ask` on the same identity resolves that
-  waiter. A `turn_id` is minted per exchange so a clarification round-trip
-  (§6.2) rejoins the right turn.
-
----
-
-## 4. Transport
-
-`bridged` is a long-lived daemon serving **multiple** concurrent clients (one primary + N
-workers), so a per-client stdio child is the wrong shape. The recommended transport is
-**streamable-HTTP / SSE** on the same bind as the REST face:
-
-```bash
-# identical on primary and every worker
-claude mcp add --transport http bridged http://127.0.0.1:8080/mcp
-```
-
-This adds an MCP-server dependency the pom does not yet carry. See [Open decisions](#10-open-decisions).
-
----
-
-## 5. Tool catalog
-
-| Tool | Caller | Blocks? | Backing (exists today?) |
-|---|---|---|---|
-| [`bridge_send`](#bridge_send) | primary | yes (default) | `Injector.enqueue` ✅ · rendezvous registry ❌ (CB-104) |
-| [`bridge_reply`](#bridge_reply) | worker | no | rendezvous ❌ · pane injection via `Injector` ✅ |
-| [`bridge_ask`](#bridge_ask) | worker | yes | reverse rendezvous ❌ |
-| [`bridge_status`](#bridge_status) | either | no | `AgentControl.status` ✅ · `Injector.activeTargets` ✅ |
-| [`bridge_spawn`](#lifecycle) | primary | no | `WorkerService.spawn` ✅ (`POST /workers`) |
-| [`bridge_list`](#lifecycle) | either | no | `WorkerService.list` ✅ (`/agents`) |
-| [`bridge_stop`](#lifecycle) | primary | no | `WorkerService.stop` ✅ (`DELETE /workers/{paneId}`) |
-| [`bridge_read`](#bridge_read) | primary | no | `AgentControl.read` ✅ |
-| [`bridge_cancel`](#bridge_cancel) | primary | no | — ❌ (future) |
-
-### Core: delegation & rendezvous
-
-#### `bridge_send`
-*(primary → worker — the headline tool, CB-104)*
-
-- **Params:** `message` (required); `target` (optional — defaults to the sole worker / default
-  profile); `timeout_seconds` (default 600); `block` (default `true`); `auto_spawn`
-  (default `true`); `turn_id` (optional — supplied when answering a worker's `bridge_ask`).
-- **Blocking (`block:true`):** enqueue `message` via the `Injector`, then hold the call open
-  until exactly one of:
-  - worker calls `bridge_reply` → `{ outcome:"reply", text }`
-  - worker calls `bridge_ask` → `{ outcome:"question", text, turn_id }`
-  - worker's `agent_status` reaches done/idle with no reply → `{ outcome:"turn_done", text:<terminal tail> }`
-  - deadline elapses → `{ outcome:"timeout" }`
-  - worker gone → error `worker_gone`
-- **Detached (`block:false`):** enqueue and return `{ outcome:"dispatched", dispatch_id }`
-  immediately. The eventual reply is injected into the primary's idle pane (§6.3), or drained
-  via `bridge_status` on a split-host primary.
-
-#### `bridge_reply`
-*(worker → primary)*
-
-- **Params:** `text` (required); `final` (default `true`).
-- **Behavior:** resolve the primary waiter registered against this worker with `text`. If no
-  waiter exists (detached delegation), `bridged` **injects the primary's idle pane** instead.
-  Returns `{ delivered:true, mode:"resolved"|"injected" }`. No `target` — identity is implicit.
-
-#### `bridge_ask`
-*(worker → primary — the reverse rendezvous)*
-
-- **Params:** `question` (required); `timeout_seconds`.
-- **Behavior:** blocks the *worker's* call. Surfaces the question to the primary (resolving its
-  open `bridge_send` with `outcome:"question"`, or injecting its pane). When the primary
-  answers — a `bridge_send` carrying the matching `turn_id` — that unblocks this call and
-  returns `{ answer }` to the worker, which continues **in the same turn**.
-
-### Worker lifecycle
-<a id="lifecycle"></a>
-Thin adapters over [`WorkerService`](1-Architecture) — parity with the existing REST routes.
-
-- **`bridge_spawn`** — `{ profile? }` → worker view (`sessionId`, `terminalId`, `paneId`,
-  `status`). Guard-checked; a boundary breach returns error `subscription_boundary` (the
-  REST `403`).
-- **`bridge_list`** — no params → all workers + `agent_status`. Read-only, either role.
-- **`bridge_stop`** — `{ target }` → tears down the pane and its dedicated tab. Idempotent.
-
-### Observability
-
-#### `bridge_status`
-*(either role — the README's 4th named tool)*
-
-- **Params:** `target?`.
-- **Behavior:** per-worker `agent_status`, queue depth (`Injector.activeTargets`), whether a
-  rendezvous is open, and ids. For the *calling* session it also reports/drains **pending
-  messages addressed to me** — the path a split-host primary's `Stop`-hook uses to wake and
-  collect replies without being injectable. Read-only, non-blocking.
-
-#### `bridge_read`
-*(primary)*
-
-- **Params:** `target`; `source` ∈ `visible | recent | recent_unwrapped | detection`.
-- **Behavior:** returns the worker's terminal text so the primary can peek at a *detached*
-  worker's progress. Adapter over `AgentControl.read`.
-
-### Control (future)
-
-#### `bridge_cancel`
-*(primary)*
-
-- **Params:** `target`. Interrupt the worker's current turn / abandon the rendezvous. No
-  backing code yet.
-
----
-
-## 6. Rendezvous flows
-
-### 6.1 Delegation — happy path
-
-One blocking call, zero polls.
+One blocking call, zero polls. The lead's call is held open by `fleetd` until the member answers.
 
 ```mermaid
 sequenceDiagram
-    participant P as Primary (Opus)
-    participant B as bridged (MCP + Injector)
+    participant P as "Lead (primary)"
+    participant B as "fleetd (MCP + Injector)"
     participant H as herdr
-    participant W as Worker (Claude)
+    participant W as "Member"
 
-    P->>B: bridge_send("do X", target=w) — blocks
-    B->>B: register waiter(w)
-    B->>H: agent.send(w, "do X")  (idle window)
-    H-->>W: prompt injected
-    W->>W: works the turn
-    W->>B: bridge_reply("result")
-    B->>B: resolve waiter(w)
-    B-->>P: { outcome:"reply", text:"result" }
+    P->>B: "fleet_send{sessionId, content} — blocks"
+    B->>B: "register waiter(sessionId)"
+    B->>H: "agent.send — only in an injectable window"
+    H-->>W: "prompt injected"
+    W->>W: "works the turn"
+    W->>B: "fleet_reply{content}"
+    B->>B: "resolve waiter"
+    B-->>P: "{ outcome: reply }"
 ```
 
-### 6.2 Clarification — reverse rendezvous (`bridge_ask`)
+**The cap that matters:** a blocking `fleet_send` is bounded by the *caller's own* MCP client
+timeout, about 60 seconds — not by the task. Anything slower than that must use the detached flow
+in §1.3, or the lead's call returns while the member is still working.
 
-The worker pauses mid-turn to ask; the primary answers; the worker resumes in the same turn.
+### 1.2 Clarification — reverse rendezvous
+
+The member pauses mid-turn to ask, the lead answers, and the member resumes **the same turn** with
+its context intact.
 
 ```mermaid
 sequenceDiagram
-    participant P as Primary
-    participant B as bridged
-    participant W as Worker
+    participant P as "Lead"
+    participant B as fleetd
+    participant W as "Member"
 
-    P->>B: bridge_send("do X", target=w) — blocks
-    B-->>W: "do X" (injected)
-    W->>B: bridge_ask("which config?") — worker blocks
-    B-->>P: { outcome:"question", text:"which config?", turn_id }
-    P->>B: bridge_send("config.yaml", target=w, turn_id) — blocks again
-    B-->>W: resolve bridge_ask → { answer:"config.yaml" }
-    W->>W: resumes same turn
-    W->>B: bridge_reply("done")
-    B-->>P: { outcome:"reply", text:"done" }
+    P->>B: "fleet_send{sessionId, content} — blocks"
+    B-->>W: "content injected"
+    W->>B: "fleet_ask{question} — member blocks"
+    B-->>P: "{ outcome: question, turnId }"
+    P->>B: "fleet_send{turnId, content} — answers THIS turn"
+    B-->>W: "fleet_ask returns the answer"
+    W->>W: "resumes the same turn"
+    W->>B: "fleet_reply{content}"
+    B-->>P: "{ outcome: reply }"
 ```
 
-### 6.3 Detached delegation — pane injection
+**Answer with `turnId`, never `sessionId`.** A `sessionId` send starts a new turn; it does not
+resolve the waiting `fleet_ask`.
 
-The primary does not block; the reply arrives later in its idle pane.
+**The window is about 55 seconds and no nudge extends it.** So never brief a member to "ask me":
+decide before delegating, or give the member an explicit default to fall back on.
+
+### 1.3 Detached delegation — the lead does not block
+
+The lead gets a ticket immediately and collects the answer later. This is the flow for any real
+task, because of the ~60s cap in §1.1.
 
 ```mermaid
 sequenceDiagram
-    participant P as Primary
-    participant B as bridged
-    participant W as Worker
+    participant P as "Lead"
+    participant B as fleetd
+    participant W as "Member"
 
-    P->>B: bridge_send("do X", target=w, block=false)
-    B-->>P: { outcome:"dispatched", dispatch_id }
-    P->>P: continues its own work
-    W->>B: bridge_reply("result")
-    Note over B: no waiter → detached path
-    B->>B: Injector.enqueue(primary_pane, "result")
-    B-->>P: injected into idle pane (status-gated)
+    P->>B: "fleet_send{sessionId, content, wait:false}"
+    B-->>P: "accepted — ticket"
+    P->>P: "continues its own work"
+    W->>B: "fleet_reply{content}"
+    Note over B: "no waiter is blocked — the reply is held"
+    B->>B: "nudge the lead's own pane (status-gated)"
+    P->>B: "fleet_poll{ticket}"
+    B-->>P: "the member's report"
+    P->>B: "fleet_ack{target, msgId}"
 ```
 
-### 6.4 Uncooperative worker — turn-done fallback
+A terminal ticket nudges the lead's pane by itself, so a detached task does not need watching. The
+nudge needs an injectable lead pane and is capped, so it is a convenience rather than a guarantee.
 
-A worker that never calls `bridge_reply` still returns a result: `bridged` reads its terminal
-tail when the turn completes.
+### 1.4 The member never replies — turn-done fallback
+
+A member that ends its turn without `fleet_reply` still produces something: `fleetd` reads its
+pane tail. This is a **fallback, not a channel** — it is lossy in three separate ways, and every
+one of them has produced a wrong answer in practice.
 
 ```mermaid
 sequenceDiagram
-    participant P as Primary
-    participant B as bridged
-    participant W as Worker
+    participant P as "Lead"
+    participant B as fleetd
+    participant W as "Member"
 
-    P->>B: bridge_send("do X", target=w) — blocks
-    B-->>W: "do X" (injected)
-    W->>W: works, never calls bridge_reply
-    B->>B: StatusPoller sees agent_status → idle/done
-    B->>B: AgentControl.read(w, "recent")
-    B-->>P: { outcome:"turn_done", text:<terminal tail> }
+    P->>B: "fleet_send — blocks or detaches"
+    B-->>W: "content injected"
+    W->>W: "works, never calls fleet_reply"
+    B->>B: "StatusPoller sees the turn end"
+    B->>B: "read the pane tail"
+    B->>B: "classify: exhausted? echoed brief? real report?"
+    B-->>P: "{ outcome: turn_done } or a named failure"
 ```
 
+The three ways it goes wrong, and what each looks like now:
+
+| What happened | What the lead used to get | What it gets today |
+|---|---|---|
+| The report is longer than the scrape window | The **end** silently cut off | Still clipped, but marked partial |
+| The member never started — spent credential | The lead's **own brief** echoed back as a report | A named failure: backend exhausted |
+| The member is simply slow | A tail of work in progress | Unchanged — read it as a hint, not a result |
+
+The echoed-brief case is the one to remember: it reads as a long, on-topic report with nothing in
+it from the member. It is suppressed now, but the general rule stands — **check the member's
+worktree with `git log` before believing a report you did not watch arrive.**
+
 ---
 
-## 7. Status gating
+## 2. Status gating
 
-Delivery only happens in a safe window. This is the state machine the `Injector` already
-enforces via `AgentStatus.injectable()`; MCP `bridge_send` is simply its producer.
+Delivery only happens in a safe window. `fleet_send` is a producer for the `Injector`, which
+already enforces this through `AgentStatus.injectable()`.
 
 ```mermaid
 stateDiagram-v2
     [*] --> IDLE
-    IDLE --> WORKING: message delivered / picks up
-    WORKING --> IDLE: turn done
-    WORKING --> BLOCKED: awaits input
-    BLOCKED --> WORKING: input delivered
-    IDLE --> UNKNOWN: detection glitch
-    BLOCKED --> UNKNOWN: detection glitch
-    UNKNOWN --> IDLE: re-detected
+    IDLE --> WORKING: "message delivered, picked up"
+    WORKING --> IDLE: "turn done"
+    WORKING --> BLOCKED: "awaits input"
+    BLOCKED --> WORKING: "input delivered"
+    IDLE --> UNKNOWN: "detection glitch"
+    BLOCKED --> UNKNOWN: "detection glitch"
+    UNKNOWN --> IDLE: "re-detected"
 
     note right of IDLE
         injectable — deliver head of FIFO
@@ -321,69 +172,17 @@ stateDiagram-v2
     end note
 ```
 
-At most one message is delivered per turn: after a send the `Injector` waits for a `WORKING`
-pickup before delivering the next, with a `PICKUP_GRACE_POLLS` fallback for turns faster than
-the poll interval. A herdr `events.subscribe` stream can later replace the sampling without
-touching this state machine.
+**At most one message per turn.** After a send, the `Injector` waits for a `WORKING` pickup before
+delivering the next, with a grace-poll fallback for turns that finish faster than the poll
+interval.
 
----
+Two consequences a lead feels directly:
 
-## 8. Error model
+- **A second send to a busy member never lands.** It reports as queued and times out. The member
+  is fine; the message simply waits, and then restarts the member when it next goes idle.
+- **A spawned member is not deliverable until it has mounted the MCP.** Until then a send waits on
+  that gate for about 60 seconds and then fails without ever reaching the pane.
 
-| Condition | `bridge_send` result | Notes |
-|---|---|---|
-| Worker replies | `{ outcome:"reply" }` | normal |
-| Worker asks | `{ outcome:"question", turn_id }` | answer with `bridge_send(turn_id)` |
-| Turn ends, no reply | `{ outcome:"turn_done" }` | terminal tail as text |
-| Deadline elapsed | `{ outcome:"timeout" }` | message may still be queued/delivered |
-| Worker vanished | error `worker_gone` | `Injector.drop` fails the queued future |
-| Guard breach on spawn | error `subscription_boundary` | REST `403` parity |
-| Delivery failed at herdr | error, message dropped | poisoned message not left blocking the FIFO |
-
-`bridge_reply` from a worker with no open waiter is **not** an error — it falls through to
-detached pane injection (§6.3).
-
----
-
-## 9. Mapping to existing code
-
-The MCP face is a thin adapter layer; nearly every capability already exists behind the REST
-seam. Only the **rendezvous registry** and the **caller-identity resolver** are new.
-
-| MCP tool | Existing collaborator | New work |
-|---|---|---|
-| `bridge_send` | `Injector.enqueue`, `AgentControl.send` | waiter registry, timeout, outcome mux (CB-104) |
-| `bridge_reply` / `bridge_ask` | `Injector` (pane injection) | reverse rendezvous, identity resolver |
-| `bridge_status` | `AgentControl.status`, `Injector.activeTargets` | pending-drain projection |
-| `bridge_spawn` / `list` / `stop` | `WorkerService.{spawn,list,stop}` | MCP adapter only |
-| `bridge_read` | `AgentControl.read` | MCP adapter only |
-
-Because the REST routes in `BridgedApp` already exercise the collaborators, MCP tools are
-validated by **parity** against those routes, not by re-testing behavior.
-
----
-
-## 10. Open decisions
-
-1. **`bridge_ask` direction.** This page defines it as *worker-asks-primary* (a genuine reverse
-   channel, matching the "inject the primary's pane" language). The alternative — a synonym for
-   a blocking primary→worker send — is weaker and produces different plumbing. **Recommend
-   worker-asks-primary.**
-2. **Detached delivery shape.** A `block:false` param on `bridge_send` (keeps the catalog
-   small) vs. a separate `bridge_dispatch` tool. **Recommend the param.**
-3. **Auto-spawn on send.** `bridge_send` provisions a worker per profile when none exists
-   (simplest primary UX) vs. requiring an explicit `bridge_spawn` first. **Recommend
-   auto-spawn, defaulting on.**
-4. **Transport & SDK.** Streamable-HTTP/SSE co-located with the REST bind (recommended) vs.
-   stdio. Requires choosing a Java MCP server SDK and adding it to the pom.
-
----
-
-## 11. Implementation staging
-
-- **CB-104** — blocking `bridge_send` + rendezvous registry + caller-identity resolver
-  (the producer that finally drives the inert `StatusPoller`).
-- **CB-1xx** — `bridge_reply` / `bridge_ask` reverse rendezvous + detached pane injection.
-- **CB-1xx** — lifecycle + observability adapters (`bridge_spawn/list/stop/status/read`).
-- **CB-1xx** — transport wiring + `claude mcp add` docs; parity tests vs. REST.
-- **Later** — `bridge_cancel`; swap `StatusPoller` for herdr `events.subscribe`.
+`UNKNOWN` is deliberately neither injectable nor a pickup. A pane whose status cannot be read is
+not a pane that is safe to write to — see fleetd #176 for what happens when a gate treats an
+unreadable pane as a ready one.

docs/RELEASE-NOTES-1.0.0.md

@@ -1,8 +1,8 @@
 # v1.0.0 — One leader, one host, complete
 
-This is the first release of **`bridged`**.
+This is the first release of **`fleetd`**.
 
-`bridged` lets one main Claude Code session (the **leader**, on your Pro/Max subscription)
+`fleetd` lets one main Claude Code session (the **leader**, on your Pro/Max subscription)
 run a team of **workers** — extra Claude Code sessions on a cheaper or local model, and
 non-Claude agents too. The leader's own session is never touched: it stays on subscription,
 with a clean environment.
@@ -19,12 +19,12 @@ of this one.
 ## One gateway for all messages
 
 - **Everyone talks through the same door.** The leader and every worker connect to the same
-  MCP server and use only its tools: `bridge_whoami` · `bridge_profiles` · `bridge_spawn` ·
-  `bridge_list` · `bridge_status` · `bridge_send` · `bridge_reply` · `bridge_ask` ·
-  `bridge_poll` · `bridge_ack` · `bridge_stop`.
+  MCP server and use only its tools: `fleet_whoami` · `fleet_profiles` · `fleet_spawn` ·
+  `fleet_list` · `fleet_status` · `fleet_send` · `fleet_reply` · `fleet_ask` ·
+  `fleet_poll` · `fleet_ack` · `fleet_stop`.
 - **You are who your connection says you are.** The bridge finds out who is calling from the
   connection itself, never from a name the caller sends. So a worker cannot pretend to be
-  someone else, and `bridge_whoami` tells each agent its own role — no guessing.
+  someone else, and `fleet_whoami` tells each agent its own role — no guessing.
 - **The subscription line cannot be crossed.** Only a spawned worker gets
   `ANTHROPIC_BASE_URL`; the leader never does. Each worker profile has a list of allowed
   model hosts, checked before anything starts.
@@ -59,7 +59,7 @@ had to ask for its replies. That gap is now closed on a single machine:
 - **The leader gets a tap on the shoulder.** When a reply lands, the bridge nudges the
   leader's own pane — only when the leader is free, and only a few times. If the leader is on
   another machine, this quietly falls back to pick-up mode; the reply still waits.
-- **Workers can ask questions.** With `bridge_ask`, a worker can pause mid-task, ask the
+- **Workers can ask questions.** With `fleet_ask`, a worker can pause mid-task, ask the
   leader something, and continue the *same* task with the answer.
 
 ## More than one kind of worker

docs/Team.md

@@ -1,9 +1,9 @@
 # Team — lead orchestrating a mixed Claude + local-LLM fleet
 
-The message server (`bridged`) delivers **one turn into one worker**. A **team** is the
+The message server (`fleetd`) delivers **one turn into one worker**. A **team** is the
 layer above it: a **Claude team-lead** that fans a job out across a **mixed fleet** of
 workers — some on Claude, some on the remote local LLM — and reduces their replies. Same
-`bridged` delivery, same subscription boundary; this doc is only about **orchestration** —
+`fleetd` delivery, same subscription boundary; this doc is only about **orchestration** —
 who the workers are, how the lead picks one, and how it runs many at once.
 
 > Delivery mechanics (blocking `POST /message`, status-gated reply envelope) live in the
@@ -12,9 +12,9 @@ who the workers are, how the lead picks one, and how it runs many at once.
 ## The team
 
 - **Team-lead** — the primary **Opus** (Claude Code, env **CLEAN**, on Pro/Max). Not a
-  worker; a **thin client of `bridged`**. It plans, routes, dispatches, and integrates, and
+  worker; a **thin client of `fleetd`**. It plans, routes, dispatches, and integrates, and
   never sets `ANTHROPIC_BASE_URL`.
-- **Workers** — a herd of `claude` panes in herdr, each an addressable `bridged` session
+- **Workers** — a herd of `claude` panes in herdr, each an addressable `fleetd` session
   with its **own model/env**:
   - **Claude workers** (clean env, e.g. Sonnet) — reasoning-heavy or high-accuracy subtasks.
   - **Local workers** (`ANTHROPIC_BASE_URL=https://llm.ltms.dev/anthropic`) — bulk, cheap, or
@@ -28,7 +28,7 @@ MCP) — only its model differs. Scale each kind horizontally by adding panes.
 ```mermaid
 flowchart TB
     LEAD["lead — Opus<br/>(Claude Code, env CLEAN)"]
-    BD["bridged<br/>message server + router"]
+    BD["fleetd<br/>message server + router"]
     HERDR["herdr<br/>panes · agent-status"]
     WC1["w-claude-1<br/>Sonnet · CLEAN"]
     WC2["w-claude-2<br/>Sonnet · CLEAN"]
@@ -62,14 +62,14 @@ flowchart TB
 | `w-local-*` | `ANTHROPIC_BASE_URL` set | local LLM | task is bulk / cheap / embarrassingly parallel |
 
 The lead applies this rubric itself, guided by its `CLAUDE.md` team charter (below). Worker
-selection is **policy in the lead**, not a `bridged` concern — `bridged` just delivers to
+selection is **policy in the lead**, not a `fleetd` concern — `fleetd` just delivers to
 the session the lead names.
 
 ## Subscription boundary in a team
 
 Unchanged from the base architecture, and it scales with the fleet: **only local-worker
 panes** launch with `ANTHROPIC_BASE_URL`. The lead and every Claude worker stay env-clean on
-the subscription. `bridged` enforces which panes may carry the off-subscription env, so
+the subscription. `fleetd` enforces which panes may carry the off-subscription env, so
 adding workers never widens the boundary.
 
 ## Parallel fan-out (map / reduce)
@@ -80,7 +80,7 @@ different workers at once, then results are gathered.
 ```mermaid
 sequenceDiagram
     participant L as lead (Opus)
-    participant B as bridged
+    participant B as fleetd
     participant WC as w-claude-1
     participant WL as w-local-1
 
@@ -100,26 +100,26 @@ sequenceDiagram
 ```
 
 - **Map:** the lead issues N concurrent blocking `POST /message` calls (one per subtask → its
-  chosen worker). Each call blocks only *that* request; `bridged` holds it open until the
+  chosen worker). Each call blocks only *that* request; `fleetd` holds it open until the
   worker's turn completes (status-gated) and returns the reply envelope.
 - **Reduce:** the lead collects the N envelopes and integrates. A slow local worker never
   blocks a fast Claude worker — wall-clock ≈ the slowest single subtask, not the sum.
 - **Detached / long jobs** use the async broker path instead of a held request (Channel 2 in
   the base architecture), so the lead never busy-polls across turns.
 
-Fan-out is bounded by the herd size (pane count) and `bridged`'s concurrency policy, not by
+Fan-out is bounded by the herd size (pane count) and `fleetd`'s concurrency policy, not by
 the lead.
 
 ## Knowing the roster
 
-The lead discovers its team from `bridged` (session list / roles) rather than hard-coding
+The lead discovers its team from `fleetd` (session list / roles) rather than hard-coding
 pane ids, so workers can be added or restarted without editing the lead. A minimal charter
 in the lead's `CLAUDE.md` turns Opus into the orchestrator:
 
 ```markdown
-## Your team (via bridged)
-You are the team-lead. Delegate through the bridged client — never launch workers yourself.
-Roster: ask bridged for current sessions/roles.
+## Your team (via fleetd)
+You are the team-lead. Delegate through the fleetd client — never launch workers yourself.
+Roster: ask fleetd for current sessions/roles.
 - w-claude-* — Claude Sonnet. Reasoning-heavy / high-accuracy subtasks.
 - w-local-*  — remote local LLM. Bulk, cheap, or parallelizable subtasks.
 
@@ -133,22 +133,22 @@ tool instead of hand-rolling the HTTP request.
 
 ## What this layer does NOT change
 
-- **Delivery** is still `bridged` → herdr `pane.send_text` + status events (Message-Server).
+- **Delivery** is still `fleetd` → herdr `pane.send_text` + status events (Message-Server).
 - **Completion timing** is still the worker status event; **reply content** still rides the
   worker `Stop`-hook envelope.
 - **Single-host** still applies: herdr's socket is local, so the whole herd lives on the
-  `bridged` host. The lead may be remote — it only needs HTTP to `bridged`.
+  `fleetd` host. The lead may be remote — it only needs HTTP to `fleetd`.
 
 ## Open questions
 
-- **Routing intelligence:** rubric-in-`CLAUDE.md` (lead decides) vs. a `bridged` role-router
+- **Routing intelligence:** rubric-in-`CLAUDE.md` (lead decides) vs. a `fleetd` role-router
   (label-based). Start with the former; promote to the latter if routing logic grows.
-- **Backpressure:** per-role concurrency caps in `bridged` so a fan-out can't exhaust the
+- **Backpressure:** per-role concurrency caps in `fleetd` so a fan-out can't exhaust the
   local gateway.
 - **Result schema:** whether reply envelopes should carry structured metadata (worker, model,
   tokens) to help the lead's reduce step.
 
 ## Status
 
-🟡 Design (2026-07-11). Orchestration layer over the selected `bridged` server; inherits
+🟡 Design (2026-07-11). Orchestration layer over the selected `fleetd` server; inherits
 herdr (chosen) + AgentAPI (fallback). Delivery unchanged — see the Message-Server design.

docs/Worker-Git-Workflow.md

@@ -20,7 +20,7 @@ requirement, not a nice-to-have.**
 - **Worktree provisioned by the daemon** — `SessionManager` creates a dedicated git worktree +
   branch per session, **hydrates it to full config parity** (below), and tears it down on release.
 - **Worker opens its own PR** — the worker commits, pushes its branch, and opens the PR/MR itself,
-  returning the PR URL in its `bridge_reply`.
+  returning the PR URL in its `fleet_reply`.
 
 ## Why worktrees (the hazard being fixed)
 
@@ -76,7 +76,7 @@ worktree checks out anyway. Amber is the real gap — untracked local config the
 3. **Never overlay the git plumbing** — the worktree's own `.git` file/branch is what gives
    isolation; that's the *one* thing that must differ from the main tree.
 
-The overlay set lives in config (`BridgedConfig.Worker.parityOverlay` — a list of repo-relative
+The overlay set lives in config (`FleetConfig.Worker.parityOverlay` — a list of repo-relative
 paths, with sane defaults) so it's auditable and per-repo tunable.
 
 > **Trust note (deliberate).** Hydrating local config means the primary's local secrets/tokens
@@ -103,7 +103,7 @@ sequenceDiagram
     Note over W: implement in the isolated worktree
     W->>G: git commit + git push (SSH, same user)
     W->>G: open PR (branch to main)
-    W-->>P: bridge_reply (prUrl, branch, summary, tests)
+    W-->>P: fleet_reply (prUrl, branch, summary, tests)
     P->>SM: release(paneId)
     SM->>G: git worktree remove wt
     Note over G: branch + PR persist for review/merge
@@ -115,9 +115,9 @@ earlier `STATE.md` idea — a PR is reviewable, mergeable, and self-describing.*
 
 ## Infra facts (verified this session)
 
-- **Remote:** `ssh://git@git.ltms.dev:2224/lms/claude-bridge.git` (gitea). Push is over **SSH** —
+- **Remote:** `ssh://git@git.ltms.dev:2224/fleet/fleetd.git` (gitea). Push is over **SSH** —
   a worker running as the same user with the same keys can `git push` **with no extra credential**.
-- **gitea is NOT in the project `.mcp.json`** (only `jetbrains`, `intellij-index`, `bridged`). The
+- **gitea is NOT in the project `.mcp.json`** (only `jetbrains`, `intellij-index`, `fleetd`). The
   primary's gitea MCP comes from a global/user config, so **workers do not inherit it**. A worker
   gets only the `bridge` MCP mounted (via `--mcp-config` launch flag).
 - **No gitea CLI** (`tea`) installed; `glab` is present but is the GitLab CLI (wrong backend).
@@ -128,7 +128,7 @@ earlier `STATE.md` idea — a PR is reviewable, mergeable, and self-describing.*
 
 | Option | Mechanism | Trade-off |
 |---|---|---|
-| **A. gitea REST + token** | Worker `curl`s `POST /api/v1/repos/lms/claude-bridge/pulls` with a scoped token injected by the daemon into the worker env | Minimal, no new server; token lives in the off-subscription worker's env (scope it tightly) |
+| **A. gitea REST + token** | Worker `curl`s `POST /api/v1/repos/fleet/fleetd/pulls` with a scoped token injected by the daemon into the worker env | Minimal, no new server; token lives in the off-subscription worker's env (scope it tightly) |
 | **B. mount gitea MCP into workers** | Add the gitea MCP to the worker's `--mcp-config` alongside `bridge` | Clean tool call, but the gitea MCP's own auth/token must be provisioned per worker; more moving parts |
 | **C. install `tea` CLI** | Worker runs `tea pr create` with a token | Another dependency to install + configure; same token question as A |
 
@@ -140,7 +140,7 @@ and the token is a single scoped secret the daemon injects like it already injec
 
 - Off-subscription workers already *could* push (SSH, same user). The **incremental grant is
   PR-create**, i.e. a gitea API token.
-- Scope the token **minimally**: the `lms/claude-bridge` repo, `write:repository` (create branch +
+- Scope the token **minimally**: the `fleet/fleetd` repo, `write:repository` (create branch +
   PR), **not** merge/admin/org. A leaked token can open PRs, not merge them — the primary/human is
   still the merge gate.
 - Inject via the daemon (env var, e.g. `GITEA_TOKEN`), never written to the worker's config dir —
@@ -153,11 +153,11 @@ and the token is a single scoped secret the daemon injects like it already injec
 |---|---|---|
 | Worktree provision/teardown | **CB-301 ext** — `SessionManager.acquire`/`release`; `WorkerSession` gains `worktree`, `branch` | daemon shells out to `git worktree add/remove` |
 | **Config-parity overlay** | **CB-301 ext** — `SessionManager.acquire`, after `git worktree add` | symlink/copy the `parityOverlay` set into the worktree so the worker is a full peer; **this is what makes worktrees viable, not a dead-end** |
-| Overlay config | `BridgedConfig.Worker.parityOverlay` — repo-relative paths, sane defaults | auditable, per-repo tunable; keep explicit + minimal (trust) |
+| Overlay config | `FleetConfig.Worker.parityOverlay` — repo-relative paths, sane defaults | auditable, per-repo tunable; keep explicit + minimal (trust) |
 | Branch naming | `worker/<ticket-slug>-<nonce>` off `main` (or a configured base) | one branch per session |
 | Commit + push + PR handoff | **CB-302** — worker-driven, guided by the skill | push = SSH; PR = option A |
 | Implementer skill | `.claude/skills/implementer/SKILL.md` | worktree-aware playbook (see below); mounts automatically since workers inherit repo cwd |
-| gitea token injection | `WorkerService` env + `BridgedConfig` | repo-scoped, minimal perms |
+| gitea token injection | `WorkerService` env + `FleetConfig` | repo-scoped, minimal perms |
 | PR review + merge | Primary (has gitea MCP + judgment) | merge on green; the human/primary gate stays |
 
 ## Implementer skill (outline)
@@ -170,7 +170,7 @@ A worker-facing playbook (sibling to the existing `reviewer` skill):
 3. **Push** your branch (`git push -u origin HEAD`).
 4. **Open a PR** to `main` (option A `curl`, or the decided mechanism) with a title/body describing
    the change and referencing the ticket.
-5. **Reply** via `bridge_reply` with the **PR URL**, branch name, files changed, and test names —
+5. **Reply** via `fleet_reply` with the **PR URL**, branch name, files changed, and test names —
    that reply is the whole handoff.
 6. Do **not** merge; do **not** touch `.mcp.json` or `wiki/`.
 

docs/Worker-Startup-and-Trust.md

@@ -1,6 +1,6 @@
 # Worker startup: working directory & the folder-trust prompt
 
-When `bridged` spawns a worker, the worker CLI may show an **interactive startup prompt** before it
+When `fleetd` spawns a worker, the worker CLI may show an **interactive startup prompt** before it
 is ready to accept a task — most importantly a *"Do you trust the files in this folder?"* dialog. An
 unattended worker parked on that prompt never becomes injectable: the status-gated injector waits for
 `idle`/`blocked`, the task is never delivered, and (worst case) a stray Enter answers the dialog
@@ -14,11 +14,11 @@ rule that **a worker inherits the primary's directory** (never `$HOME`), and how
 
 ```mermaid
 flowchart TD
-    A["bridge_spawn / POST /workers"] --> B{"explicit cwd?<br/>(profile cwd or spawn arg)"}
+    A["fleet_spawn / POST /workers"] --> B{"explicit cwd?<br/>(profile cwd or spawn arg)"}
     B -->|"yes — told otherwise"| C["use that cwd"]
     B -->|"no"| D{"caller PID resolvable?<br/>(MCP peer PID)"}
     D -->|"yes"| E["cwd = the primary's cwd<br/>lsof -a -p PID -d cwd"]
-    D -->|"no (REST / off-host)"| F["cwd = bridged daemon cwd<br/>(never $HOME by assumption)"]
+    D -->|"no (REST / off-host)"| F["cwd = fleetd daemon cwd<br/>(never $HOME by assumption)"]
     C --> G["ensureWorkspace → tab.create → agent.start {cwd}"]
     E --> G
     F --> G
@@ -51,10 +51,10 @@ only affect the seed shell, which the bridge closes).
 | # | Source | When |
 |---|--------|------|
 | 1 | Explicit `cwd` — a per-profile `cwd:` in config, or a spawn argument | "told otherwise" — pin a fixed workdir |
-| 2 | The **primary's cwd**, auto-detected from the `bridge_spawn` caller | normal MCP spawn from the primary |
-| 3 | The `bridged` daemon's own cwd | REST spawn / off-host caller — **never `$HOME`** |
+| 2 | The **primary's cwd**, auto-detected from the `fleet_spawn` caller | normal MCP spawn from the primary |
+| 3 | The `fleetd` daemon's own cwd | REST spawn / off-host caller — **never `$HOME`** |
 
-The primary's cwd (source 2) is discoverable with no new plumbing: `bridged` already resolves the MCP
+The primary's cwd (source 2) is discoverable with no new plumbing: `fleetd` already resolves the MCP
 caller's loopback **peer PID** for connection identity (`ConnectionIdentity` → `LsofPeerPidLookup`);
 the same PID yields its cwd via `lsof -a -p <pid> -d cwd -Fn` (the `n…` line). The primary maps to no
 worker pane (it is not a worker), but its PID and cwd are still readable.
@@ -62,10 +62,10 @@ worker pane (it is not a worker), but its PID and cwd are still readable.
 ```mermaid
 sequenceDiagram
     participant P as "Primary (main)"
-    participant B as "bridged"
+    participant B as "fleetd"
     participant O as "OS (lsof)"
     participant H as "herdr"
-    P->>B: "bridge_spawn {profile} (no cwd)"
+    P->>B: "fleet_spawn {profile} (no cwd)"
     B->>O: "peer PID for this connection's port"
     O-->>B: "pid"
     B->>O: "cwd of pid (lsof -d cwd)"
@@ -77,9 +77,9 @@ sequenceDiagram
 
 *Figure 2 — a no-cwd spawn inherits the primary's directory from the caller's PID.*
 
-> **Status:** implemented (CB-112). `bridged` threads the resolved `cwd` onto **`agent.start {cwd}`**
+> **Status:** implemented (CB-112). `fleetd` threads the resolved `cwd` onto **`agent.start {cwd}`**
 > (verified: the worker process is rooted there), keeping the single shared worker space. On an MCP
-> `bridge_spawn` the primary's cwd is auto-detected from the caller's PID; over REST (no MCP caller)
+> `fleet_spawn` the primary's cwd is auto-detected from the caller's PID; over REST (no MCP caller)
 > it is the explicit `cwd` param else the daemon's cwd. Both placements (`tab` and legacy `pane`)
 > carry it, since it rides `agent.start`.
 
@@ -133,5 +133,5 @@ unattended.
 
 ## See also
 
-- `docs/MCP-Contract.md` — the tool surface (`bridge_spawn`, `bridge_profiles`, …).
+- `docs/MCP-Contract.md` — the tool surface (`fleet_spawn`, `fleet_profiles`, …).
 - `wiki/2-Message-Server.md` — the herdr `agent.*` / `workspace.*` schema (`workspace.create {cwd}`).

docs/wiki-audit.md

@@ -0,0 +1,208 @@
+# Wiki audit for #168
+
+**Source checked:** `.wiki-snapshot/` at `68e32c6` (2026-08-31). I did not use
+`wiki/`. Code references below are from the current `fleetd` source tree. A quoted
+line is a concrete claim that needs correction, unless the table says `KEEP`.
+
+| Page | Verdict | One-line reason |
+|---|---|---|
+| `Home.md` | REVISE | Good overview, but it still names the retired product. |
+| `_Sidebar.md` | REVISE | The heading still says `claude-bridge`. |
+| `1-Architecture.md` | REBUILD | Its component contract mixes current names with removed tools, routes, and planned backends. |
+| `2-Message-Server.md` | REBUILD | The claimed MCP schema, mount command, REST/SSE surface, and fallback paths are pre-build design. |
+| `3-Approaches.md` | REVISE | Useful research history, but it presents unbuilt AgentAPI as a selectable fallback. |
+| `4-Setup.md` | RETIRE | It is an intentional stub that only redirects to chapter 13. |
+| `5-Operations.md` | RETIRE | It is an intentional stub that only redirects to chapter 13. |
+| `6-Team.md` | REBUILD | It teaches role-addressed sends and a Claude-only team model that the shipped API does not have. |
+| `7-Use-Cases.md` | REBUILD | Its flagship flow depends on removed `ccs` profiles and removed send parameters. |
+| `8-Roadmap.md` | REBUILD | It is a historical plan, but it presents old implementation choices and planned work as the current stack. |
+| `9-Implementation.md` | REBUILD | Its package, class, endpoint, and outcome map has drifted from the source. |
+| `10-Cross-Host-Messaging.md` | REVISE | It labels most federation work proposed, but misses the shipped `coordinator:` lead channel. |
+| `11-Features.md` | REVISE | It is the right catalogue, but code-path names are old and it misses the second-herdr-daemon capability. |
+| `12-Claude-to-OpenCode.md` | REVISE | The porting guide is mostly current, but calls the product and spawned-member path a bridge. |
+| `13-User-Guide.md` | REVISE | It is the best operator page, but needs the product rename and the second-herdr-daemon setup. |
+
+## Pages needing work
+
+### `Home.md` — REVISE
+
+- Quote: `# claude-bridge` (line 1) and `` `claude-bridge` keeps`` (line 11).
+  The product is `fleet` / `fleetd`. The MCP server identifies itself as `fleet` in
+  `fleetd/src/main/java/dev/ltms/fleet/mcp/FleetMcp.java:313-315`.
+- Quote: `AgentAPI ... swappable fallback injector` (lines 73-76).
+  There is no AgentAPI implementation under `fleetd/src/main/java`; the actual
+  launchers are selected by `Profile.kind` in
+  `fleetd/src/main/java/dev/ltms/fleet/config/FleetConfig.java:265-270`.
+
+### `_Sidebar.md` — REVISE
+
+- Quote: `### 📖 claude-bridge` (line 1).
+  Rename it to `fleet`. `FleetMcp` registers the current product-facing tool set at
+  `fleetd/src/main/java/dev/ltms/fleet/mcp/FleetMcp.java:301-326`.
+
+### `1-Architecture.md` — REBUILD
+
+- Quote: `` `claude-bridge` lets`` (line 3). The product was renamed; the MCP
+  server name is `fleet` (`FleetMcp.java:313-315`).
+- Quote: ``fleet_read`` in the tool list (line 102). No such tool is registered.
+  The complete registered list is `fleet_send` through `fleet_whoami` at
+  `FleetMcp.java:301-326`; `fleet_read` is absent.
+- Quote: `SSE (GET /events)` (line 143). `FleetApp.build()` registers no `/events`
+  route; its routes are listed at `FleetApp.java:143-159`.
+- Quote: `Redis Streams / NATS JetStream, or an embedded queue` (line 106).
+  The shipped durable inbox is AMQP, configured by `broker`, at
+  `FleetConfig.java:49-50` and `FleetConfig.java:655-714`.
+- Quote: `AgentAPI (fallback)` (line 107). No AgentAPI adapter exists; shipped
+  launcher kinds are `claude-code` and `opencode` (`FleetConfig.java:265-270`).
+
+### `2-Message-Server.md` — REBUILD
+
+- Quote: `claude mcp add --transport http bridge http://127.0.0.1:8080/mcp`
+  (line 67). The daemon defaults to port `8765` in `FleetConfig.java:183-187`,
+  and identifies its server as `fleet` at `FleetMcp.java:313-315`.
+- Quote: ``fleet_send(message, target?, {block, timeout_seconds, auto_spawn,
+  turn_id})`` (line 80). The real parameters are `sessionId`, `content`,
+  `timeoutMs`, `wait`, `turnId`, and `coordId` (`FleetMcp.java:1096-1108`).
+- Quote: ``fleet_read(target, source)`` (line 85). It is not registered; see the
+  complete registration at `FleetMcp.java:301-326`.
+- Quote: `docs/MCP-Contract.md ... normative` (lines 87-88). That is not a valid
+  reference: only §6 is current, as the current operator guide itself says at
+  `.wiki-snapshot/13-User-Guide.md:466`.
+- Quote: `SSE (GET /events)` (line 45). No route exists in the built REST surface,
+  `FleetApp.java:143-159`.
+
+### `3-Approaches.md` — REVISE
+
+- Quote: `AgentAPI ... remains a swappable fallback injector` (lines 78-84).
+  It was never built. The shipped adapter selection is only `claude-code` or
+  `opencode` (`FleetConfig.java:265-270`). Keep it as discarded research, not an
+  operational fallback.
+- Quote: `claude-bridge` (line 109). Rename the product to `fleet`; the runtime
+  package is `dev.ltms.fleet`, for example `FleetMcp.java:1`.
+
+### `4-Setup.md` — RETIRE
+
+It is a 25-line redirect and says its procedure was never written (lines 3-9).
+Chapter 13 is the maintained install procedure. Keeping a second navigation page
+adds no working documentation.
+
+### `5-Operations.md` — RETIRE
+
+It is a 35-line redirect and says its runbook was never written (lines 3-14).
+Chapter 13 now owns run and recovery instructions.
+
+### `6-Team.md` — REBUILD
+
+- Quote: `fleet_send {role: w-claude, prompt: A}` (line 98). `fleet_send` accepts
+  `sessionId` and `content`, not `role` or `prompt` (`FleetMcp.java:1096-1108`).
+- Quote: `some on Claude, some on the remote local LLM` (lines 3-5) and `Every
+  worker is ... Claude Code` (line 25). `opencode` is a first-class launcher kind,
+  not a Claude worker (`FleetConfig.java:265-270`).
+- Quote: `fleetd's concurrency policy` (line 121). The configured capacity control
+  is per-profile `maxLoad` (`FleetConfig.java:251-264`), not the role routing model
+  described here.
+
+### `7-Use-Cases.md` — REBUILD
+
+- Quote: `ccs profile` (line 10), `ccs + herdr` (line 22), and `ccs-spawn`
+  (line 45). The configuration has `profiles` and `fleet`, not `ccs`:
+  `FleetConfig.java:34-58` and `FleetConfig.java:81-101`.
+- Quote: `fleet_send({"to", "kind", "body", "block"})` (lines 55-62).
+  None of those are the shipped send parameters. The schema is
+  `FleetMcp.java:1096-1108`.
+- Quote: `fleet_list() → { "profiles": ... }` (lines 74-80). `fleet_list` is a
+  roster view; `fleet_profiles` is the configured-backend view, as registered at
+  `FleetMcp.java:307-311` and described at `FleetMcp.java:1176-1182`.
+
+### `8-Roadmap.md` — REBUILD
+
+- Quote: `Java 21+` (line 43). The current project guidance and source use Java 25;
+  the `FleetConfig` source itself uses Java 25 unnamed lambda parameters, for
+  example `FleetConfig.java:102`.
+- Quote: `herdr 0.7.0 / protocol 14` (line 46). The current REST health endpoint
+  reports the live protocol returned by herdr (`FleetApp.java:240-244`), while the
+  current operator guide records protocol 19 at
+  `.wiki-snapshot/13-User-Guide.md:76-85`.
+- Quote: `ccs <profile> claude` and `ccs env <profile>` (lines 47-48). Shipped
+  configuration uses `Profile` records and launcher `kind`,
+  `FleetConfig.java:313-330` and `FleetConfig.java:265-270`.
+- Quote: `Redis Streams via Lettuce` (line 50). The actual durable inbox is AMQP
+  `broker`, `FleetConfig.java:655-714`.
+
+### `9-Implementation.md` — REBUILD
+
+- Quote: `rest.FleetdApp` and `mcp.BridgeMcp` (lines 29-30). The classes are
+  `rest.FleetApp` and `mcp.FleetMcp` (`FleetApp.java:46`; `FleetMcp.java:67`).
+- Quote: `dev.ltms.fleetd` (line 67). The source package is `dev.ltms.fleet`
+  (`FleetMcp.java:1`).
+- Quote: `WorkerPresence` (line 110). The current class is `MemberPresence`, as
+  imported and used by `FleetMcp` at `FleetMcp.java:12` and `465-469`.
+- Quote: the outcome list ending in `STALE_TURN` (lines 128-131). The code also
+  has `BACKEND_EXHAUSTED` (`FleetMcp.java:550-554`) and async `ASKING` handling
+  (`FleetMcp.java:664-668`).
+- Quote: `FleetdApp` (line 207) and `FleetdConfig` (line 211). These names do not
+  resolve; current classes are `FleetApp` and `FleetConfig`.
+
+### `10-Cross-Host-Messaging.md` — REVISE
+
+- Quote: the chapter says the cross-host fabric is proposed except for the
+  single-host inbox (lines 3-8). Cross-host **lead-to-lead** delivery shipped:
+  `fleet_send` accepts `coordId` (`FleetMcp.java:1094-1107`) and publishes it at
+  `FleetMcp.java:616-641`; configuration has `coordinator` at
+  `FleetConfig.java:74-78` and `99-101`.
+- Quote: `bridge.dlx` (line 90). This product name is stale. The shipped lead path
+  uses `LeadChannel`, not the proposed exchange flow (`FleetMcp.java:95-96` and
+  `616-641`). Keep the proposed federation design, but add a clear shipped/proposed
+  boundary for CB-637.
+
+### `11-Features.md` — REVISE
+
+- Quote: `mcp/BridgeMcp` (line 22), `config/FleetdConfig` (lines 25-27), and other
+  index references. These paths no longer resolve; the source classes are
+  `mcp/FleetMcp` (`FleetMcp.java:67`) and `config/FleetConfig`
+  (`FleetConfig.java:81`).
+- Quote: `fleet_whoami` returns only `primary` or `worker` (lines 99-100).
+  It also returns `architect` (`FleetMcp.java:1235-1244`).
+- The page needs the missing separate member-herdr-daemon feature listed below.
+
+### `12-Claude-to-OpenCode.md` — REVISE
+
+- Quote: `same bridge mount` (line 5) and `a bridge-spawned worker` (line 94).
+  Rename the product path to `fleet`. The daemon exposes the MCP server as `fleet`
+  (`FleetMcp.java:313-315`), and profiles select OpenCode with `kind: opencode`
+  (`FleetConfig.java:332-335`).
+- Quote: the sample mount name is `fleetd` (line 67). The server name is `fleet`;
+  update the sample to avoid teaching a second product name.
+
+### `13-User-Guide.md` — REVISE
+
+- Quote: `The bridge is the only channel` (line 63). The invariant is correct, but
+  the product term needs the `fleet` rename. The daemon's MCP server name is
+  `fleet` (`FleetMcp.java:313-315`).
+- Quote: it describes one herdr socket (lines 72-85). It needs the optional
+  `memberHerdrSocket` setup and two-daemon health meaning. The config key is in
+  `FleetConfig.java:34-37`, and `/healthz` checks both daemons when configured at
+  `FleetApp.java:210-245`.
+
+## MISSING
+
+`11-Features.md` has a body section for **routing members through a separate herdr daemon**
+(`## memberHerdrSocket`, line 2174), but **no row in the index table** at the top of the page
+(lines 20-95). That table is how the page is meant to be read, so a capability absent from it is
+effectively undiscoverable. Lead note: this is my own omission — I added the section on 2026-08-31
+and did not add the matching row. Fixed in the wiki at `68e32c6`'s successor.
+
+The original audit stated the feature had no entry at all. That was wrong: the section exists. The
+gap is the index row. Recorded here rather than silently corrected, because the difference matters —
+"undocumented" and "documented but unindexed" are different jobs.
+
+Evidence for the feature itself: `FleetConfig.java:34-37` and `FleetApp.java:103-115`, `210-245`,
+and `247-263`.
+
+## Audit method and coverage
+
+I checked all 15 pages. I checked concrete tool, route, config, class, file, and
+product-name claims claim-by-claim on 11 pages: Home, Sidebar, 1, 2, 4, 5, 6, 7, 9,
+11, and 13. I skimmed the remaining four long historical or research pages (3, 8, 10,
+12), then checked their concrete claims that affect the verdict. This is an audit of
+the supplied snapshot, not a wiki rewrite.

e2e/README.md

@@ -2,11 +2,11 @@
 
 A standard, repeatable **live** end-to-end test of the two-way channel: it drives a real
 multi-turn conversation between a primary and an off-subscription worker **through the
-running `bridged` daemon**, captures the full transcript, and grades the channel.
+running `fleetd` daemon**, captures the full transcript, and grades the channel.
 
 This is the committed form of the ad-hoc channel test that discovered the CB-115 gaps
 (herdr `unknown` misclassification wedging delivery, dirty completion scrapes, and workers
-never calling `bridge_reply` in conversation). Run it after any change to the injector,
+never calling `fleet_reply` in conversation). Run it after any change to the injector,
 status handling, completion/failure paths, or the worker reply charter.
 
 ## What it exercises
@@ -20,7 +20,7 @@ construction** — it only calls the bridge's loopback REST face.
 ```mermaid
 sequenceDiagram
     participant T as conversation_test.py
-    participant B as bridged (REST)
+    participant B as fleetd (REST)
     participant W as worker (off-sub)
     T->>B: POST /workers  (spawn)
     T->>B: GET /sessions/{id}/status  (await ready)
@@ -28,7 +28,7 @@ sequenceDiagram
         T->>B: POST /sessions/{id}/message {wait:false}
         B-->>T: ticket
         B->>W: inject prompt (status-gated)
-        W-->>B: bridge_reply
+        W-->>B: fleet_reply
         T->>B: GET /tasks/{ticket}  (poll)
         B-->>T: done + reply
     end
@@ -37,7 +37,7 @@ sequenceDiagram
 
 ## Prerequisites
 
-- `bridged` is running (default REST on `http://127.0.0.1:8765`) with at least one worker
+- `fleetd` is running (default REST on `http://127.0.0.1:8765`) with at least one worker
   profile configured and its backend reachable.
 - herdr is up (the daemon needs it).
 - Python 3 (standard library only — no pip installs).
@@ -68,7 +68,7 @@ Per-turn grade:
 
 | Grade      | Meaning                                                              |
 |------------|---------------------------------------------------------------------|
-| `OK`       | delivered and resolved by an explicit `bridge_reply` (`source=reply`) |
+| `OK`       | delivered and resolved by an explicit `fleet_reply` (`source=reply`) |
 | `DEGRADED` | delivered and answered, but resolved via completion-scrape fallback |
 | `EMPTY`    | turn completed but the reply was empty                              |
 | `FAILED`   | the worker's turn ended in failure (`phase=failed`)                 |

e2e/conversation_sustained_test.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 """Sustained back-and-forth bridge test — ONE primary, ONE worker, many dependent turns
-over a fixed wall-clock window (default 5 minutes), through the running `bridged` daemon.
+over a fixed wall-clock window (default 5 minutes), through the running `fleetd` daemon.
 
 Where conversation_test.py proves a handful of turns work and issue_hunt_test.py proves
 fan-out isolation, this proves the channel stays healthy under a *sustained, stateful*
@@ -24,7 +24,7 @@ Usage:
     --turn-timeout  per-turn max wait, seconds    (default 150)
     --keep-worker   do not stop the worker at the end
 
-Exit code: 0 if every turn in the window resolved via a clean bridge_reply with no channel
+Exit code: 0 if every turn in the window resolved via a clean fleet_reply with no channel
 break; 1 otherwise. A live per-turn log streams to stdout so the run can be watched.
 """
 import argparse
@@ -47,12 +47,12 @@ STEPS = [7, 3, 11, 5, 9, 4, 13, 6, 8, 2]
 RULES = (
     "Let's play a running-total game across several messages. The total starts at 0. "
     "In each message I'll tell you to add a number; keep the running total yourself and "
-    "reply via bridge_reply with ONLY the current total as a plain integer — no words, no "
+    "reply via fleet_reply with ONLY the current total as a plain integer — no words, no "
     "punctuation, just the number. Do not restate the arithmetic. First move: add {step}."
 )
-NEXT = ("Add {step}. Reply via bridge_reply with only the new running total.")
+NEXT = ("Add {step}. Reply via fleet_reply with only the new running total.")
 REANCHOR = ("Let's re-sync — the running total is {total}. Now add {step}. Reply via "
-            "bridge_reply with only the new running total.")
+            "fleet_reply with only the new running total.")
 
 
 def parse_int(reply):
@@ -142,15 +142,15 @@ def main():
     print("=" * 72)
     print(f"SUSTAINED CONVERSATION SUMMARY — 1 primary <-> 1 worker over {dur}s (~{dur/60:.1f} min)")
     print(f"  turns:            {turns}")
-    print(f"  clean bridge_reply exchanges: {oks + drifts}/{turns}   (channel breaks: {breaks})")
+    print(f"  clean fleet_reply exchanges: {oks + drifts}/{turns}   (channel breaks: {breaks})")
     print(f"  arithmetic correct (continuity held): {oks}/{turns}   (drifts: {drifts})")
     print(f"  latency:          avg {avg}s over {turns} turns")
     ok = breaks == 0 and turns >= 2
     if ok and drifts == 0:
-        print("  RESULT: PASS — every turn resolved via bridge_reply and the worker held the "
+        print("  RESULT: PASS — every turn resolved via fleet_reply and the worker held the "
               "running total across the whole window.")
     elif ok:
-        print(f"  RESULT: PASS (channel) — every turn resolved via bridge_reply for the full "
+        print(f"  RESULT: PASS (channel) — every turn resolved via fleet_reply for the full "
               f"window; {drifts} arithmetic drift(s) (worker recovered after re-anchor).")
     else:
         print("  RESULT: FAIL — the channel broke on at least one turn (see CHANNEL BREAK above).")

e2e/conversation_test.py

@@ -1,10 +1,10 @@
 #!/usr/bin/env python3
 """Standard bridge conversation test — a multi-turn primary↔worker exchange through
-the running `bridged` daemon, fully captured, with automatic gap analysis.
+the running `fleetd` daemon, fully captured, with automatic gap analysis.
 
 This is the repeatable form of the ad-hoc channel test that surfaced the CB-115 gaps
 (herdr `unknown` misclassification, dirty completion scrape, workers not calling
-bridge_reply). It drives a real off-subscription worker over the live gateway exactly
+fleet_reply). It drives a real off-subscription worker over the live gateway exactly
 as a primary Opus session would (async fire-and-poll), records every turn, and grades
 the channel.
 
@@ -130,9 +130,9 @@ def grade(rec):
     phase, source, reply = rec["phase"], rec["source"], rec["reply"]
     has_reply = bool(reply and reply.strip())
     if phase == "done" and source == "reply" and has_reply:
-        return "OK", "clean explicit bridge_reply"
+        return "OK", "clean explicit fleet_reply"
     if phase == "done" and has_reply:
-        return "DEGRADED", f"resolved via {source} (worker did not call bridge_reply)"
+        return "DEGRADED", f"resolved via {source} (worker did not call fleet_reply)"
     if phase == "done" and not has_reply:
         return "EMPTY", "turn completed but reply was empty"
     if phase == "failed":
@@ -206,7 +206,7 @@ def main():
     ok = all(g in ("OK", "DEGRADED") for g in grades)
     reply_clean = all(g == "OK" for g in grades)
     if reply_clean:
-        print("  RESULT: PASS — every turn delivered and got a clean bridge_reply.")
+        print("  RESULT: PASS — every turn delivered and got a clean fleet_reply.")
     elif ok:
         print("  RESULT: PASS (with notes) — every turn delivered & replied, but some via fallback.")
     else:

e2e/fleet_ask_test.py

@@ -1,16 +1,16 @@
 #!/usr/bin/env python3
-"""Live bridge_ask test — the REVERSE rendezvous (CB-205), watched end to end.
+"""Live fleet_ask test — the REVERSE rendezvous (CB-205), watched end to end.
 
-Every other harness drives the forward path: primary `bridge_send` → worker `bridge_reply`.
+Every other harness drives the forward path: primary `fleet_send` → worker `fleet_reply`.
 This drives the one that runs the other way. A worker is told to pause its delegated turn,
-ask the primary a question via `bridge_ask`, and only finish once it has the answer — so the
+ask the primary a question via `fleet_ask`, and only finish once it has the answer — so the
 turn round-trips primary→worker→primary→worker inside a SINGLE delegation.
 
 The mechanics that only this path exercises:
 
   • a worker's mid-turn question surfacing on the primary's *own* blocked send (Outcome.QUESTION),
   • the `turnId` correlation that lets the primary answer the exact paused turn,
-  • the answer resuming that same turn and the worker's final `bridge_reply` landing on the
+  • the answer resuming that same turn and the worker's final `fleet_reply` landing on the
     re-opened forward waiter (never a stale or cross-wired one).
 
 It is two blocking REST calls, no polling:
@@ -24,7 +24,7 @@ Like the rest of the suite it talks ONLY to the bridge's REST face on loopback
 ANTHROPIC_BASE_URL and never touches herdr, so it is subscription-safe by construction.
 
 Usage:
-    python3 bridge_ask_test.py [--base URL] [--profile NAME] [--repo DIR] [--out DIR]
+    python3 fleet_ask_test.py [--base URL] [--profile NAME] [--repo DIR] [--out DIR]
                                [--send-timeout SECS] [--answer-timeout SECS] [--keep-worker]
 
     --base            bridge REST base URL              (default http://127.0.0.1:8765)
@@ -57,14 +57,14 @@ ANSWER_COLOR = "blue"
 
 # A task that CANNOT be completed without asking: the worker is not told which color to choose,
 # only that the primary will name one when asked. So a correct final reply is only reachable by
-# actually calling bridge_ask and using the answer.
+# actually calling fleet_ask and using the answer.
 TASK_PROMPT = (
     "You are a bridge worker in a quick coordination game. You do NOT know which color to pick — "
     "only the primary does. Do exactly this, in order:\n"
-    "1. Call the `bridge_ask` tool with EXACTLY this question: \"PICK A COLOR: red or blue?\"\n"
+    "1. Call the `fleet_ask` tool with EXACTLY this question: \"PICK A COLOR: red or blue?\"\n"
     "2. The primary will answer with one color word. Take that color and uppercase it.\n"
-    "3. Call `bridge_reply` with EXACTLY one line: CHOSEN=<COLOR>  (e.g. CHOSEN=GREEN if told green).\n"
-    "Do not guess a color. Do not call bridge_reply before bridge_ask has returned an answer. "
+    "3. Call `fleet_reply` with EXACTLY one line: CHOSEN=<COLOR>  (e.g. CHOSEN=GREEN if told green).\n"
+    "Do not guess a color. Do not call fleet_reply before fleet_ask has returned an answer. "
     "Do nothing else — no file reads, no other tools."
 )
 
@@ -134,7 +134,7 @@ def run(base, profile, repo, send_timeout, answer_timeout):
     rec.update(tid=tid, pane=pane, spawned=True)
     await_ready(base, tid)
 
-    # 1) Delegate the ask-forcing task. This blocks until the worker calls bridge_ask, at which
+    # 1) Delegate the ask-forcing task. This blocks until the worker calls fleet_ask, at which
     #    point our own send unblocks carrying the question and the turnId to answer on.
     print(f"[{now()}] delegating task (blocks until the worker asks; up to {send_timeout}s)…")
     t0 = time.time()
@@ -159,7 +159,7 @@ def run(base, profile, repo, send_timeout, answer_timeout):
         rec.update(phase="no_turnid", detail="question surfaced without a turnId to answer on")
         return rec
 
-    # 2) Answer on that exact turn. This blocks again until the resumed worker calls bridge_reply.
+    # 2) Answer on that exact turn. This blocks again until the resumed worker calls fleet_reply.
     print(f"[{now()}] answering '{ANSWER_COLOR}' on turn {rec['turnId']} (blocks until reply; up to {answer_timeout}s)…")
     t1 = time.time()
     rec["phase"] = "awaiting_reply"
@@ -188,7 +188,7 @@ def run(base, profile, repo, send_timeout, answer_timeout):
 def grade(rec):
     """PASS only if the worker asked, the turn resumed, and the reply reflects the answer."""
     if rec["phase"] == "no_question":
-        return "NO_ASK", "the worker finished/stalled without ever calling bridge_ask"
+        return "NO_ASK", "the worker finished/stalled without ever calling fleet_ask"
     if rec["phase"] in ("send_error", "answer_error", "spawn"):
         return "ERROR", rec.get("detail") or "transport error before the round-trip completed"
     if rec["phase"] == "no_turnid":
@@ -203,26 +203,26 @@ def grade(rec):
             return "OK", "asked, resumed the same turn, and the reply reflected the primary's answer"
         if reflected:
             return "DEGRADED", f"reply reflected the answer but resolved via {rec['replySource']} " \
-                               "(worker did not call bridge_reply cleanly)"
+                               "(worker did not call fleet_reply cleanly)"
         return "WRONG_ANSWER", f"the worker replied but did not reflect '{ANSWER_COLOR}' — " \
                                f"the answer may not have reached the resumed turn: {rec['reply']!r}"
     return "WEDGE", f"unexpected terminal phase {rec['phase']}: {rec.get('detail')}"
 
 
 def write_transcript(out_dir, rec, meta):
-    path = out_dir / "bridge_ask_transcript.md"
+    path = out_dir / "fleet_ask_transcript.md"
     g, note = grade(rec)
     with path.open("w") as f:
-        f.write(f"# Live bridge_ask — reverse rendezvous — {datetime.now():%Y-%m-%d %H:%M}\n\n")
+        f.write(f"# Live fleet_ask — reverse rendezvous — {datetime.now():%Y-%m-%d %H:%M}\n\n")
         f.write(f"One worker paused its delegated turn to ask the primary, then resumed with the "
                 f"answer (profile `{meta['profile']}`). Result: **`{g}`**.\n\n")
         f.write("## Round-trip\n\n")
         f.write(f"1. **primary → worker** (delegation): the ask-forcing task.\n")
-        f.write(f"2. **worker → primary** (`bridge_ask`, {rec.get('ask_latency')}s): "
+        f.write(f"2. **worker → primary** (`fleet_ask`, {rec.get('ask_latency')}s): "
                 f"{rec.get('question')!r} — surfaced on the primary's blocked send as a "
                 f"`question` with `turnId={rec.get('turnId')}`.\n")
         f.write(f"3. **primary → worker** (answer on that turn): `{ANSWER_COLOR}`.\n")
-        f.write(f"4. **worker → primary** (`bridge_reply`, {rec.get('answer_latency')}s, "
+        f.write(f"4. **worker → primary** (`fleet_reply`, {rec.get('answer_latency')}s, "
                 f"source={rec.get('replySource')}): {rec.get('reply')!r}\n\n")
         f.write(f"> **{g}:** {note}\n")
         if rec.get("detail"):
@@ -231,7 +231,7 @@ def write_transcript(out_dir, rec, meta):
 
 
 def main():
-    ap = argparse.ArgumentParser(description="Live bridge_ask reverse-rendezvous test (CB-205)")
+    ap = argparse.ArgumentParser(description="Live fleet_ask reverse-rendezvous test (CB-205)")
     ap.add_argument("--base", default="http://127.0.0.1:8765")
     ap.add_argument("--profile", default=None)
     ap.add_argument("--repo", default=str(REPO_ROOT))
@@ -241,7 +241,7 @@ def main():
     ap.add_argument("--keep-worker", action="store_true")
     args = ap.parse_args()
 
-    print(f"[{now()}] live bridge_ask: 1 primary, 1 worker "
+    print(f"[{now()}] live fleet_ask: 1 primary, 1 worker "
           f"(profile={args.profile or 'default'}, repo={args.repo})\n")
 
     rec = {"spawned": False, "pane": None}
@@ -262,7 +262,7 @@ def main():
 
     print()
     print("=" * 72)
-    print("LIVE bridge_ask SUMMARY — reverse rendezvous (CB-205)")
+    print("LIVE fleet_ask SUMMARY — reverse rendezvous (CB-205)")
     print(f"  asked:    {rec.get('question')!r}  (turnId={rec.get('turnId')}, {rec.get('ask_latency')}s)")
     print(f"  answered: {ANSWER_COLOR!r}")
     print(f"  replied:  {rec.get('reply')!r}  (source={rec.get('replySource')}, {rec.get('answer_latency')}s)")

e2e/fleet_ask_transcript.md

@@ -1,12 +1,12 @@
-# Live bridge_ask — reverse rendezvous — 2026-07-16 16:30
+# Live fleet_ask — reverse rendezvous — 2026-07-16 16:30
 
 One worker paused its delegated turn to ask the primary, then resumed with the answer (profile `default`). Result: **`OK`**.
 
 ## Round-trip
 
 1. **primary → worker** (delegation): the ask-forcing task.
-2. **worker → primary** (`bridge_ask`, 6.6s): 'PICK A COLOR: red or blue?' — surfaced on the primary's blocked send as a `question` with `turnId=term_656bb47d2c42a9e#1`.
+2. **worker → primary** (`fleet_ask`, 6.6s): 'PICK A COLOR: red or blue?' — surfaced on the primary's blocked send as a `question` with `turnId=term_656bb47d2c42a9e#1`.
 3. **primary → worker** (answer on that turn): `blue`.
-4. **worker → primary** (`bridge_reply`, 7.9s, source=reply): 'CHOSEN=BLUE'
+4. **worker → primary** (`fleet_reply`, 7.9s, source=reply): 'CHOSEN=BLUE'
 
 > **OK:** asked, resumed the same turn, and the reply reflected the primary's answer

e2e/issue_hunt_test.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 """Standard bridge fan-out test — ONE primary vs MANY workers, concurrently, for
-issue hunting through the running `bridged` daemon, fully captured, with gap analysis.
+issue hunting through the running `fleetd` daemon, fully captured, with gap analysis.
 
 Where conversation_test.py exercises a single worker over multiple turns, this drives
 the path that only appears under fan-out: the primary spawns N workers, sends each a
@@ -12,7 +12,7 @@ and collects every reply concurrently. That stresses what a single worker never
   • reply routing under concurrency (worker A's answer must never resolve worker B's send),
 
 and, as the payload, whether a fleet of off-subscription workers can actually surface
-real issues in the repo and report them back structurally via bridge_reply.
+real issues in the repo and report them back structurally via fleet_reply.
 
 It talks ONLY to the bridge's REST face on loopback — it never sets ANTHROPIC_BASE_URL
 and never touches herdr directly, so it is subscription-safe by construction.
@@ -53,18 +53,18 @@ REPO_ROOT = HERE.parent
 # hot files this project has been iterating on, so a real issue is plausible to find.
 DEFAULT_ASSIGNMENTS = [
     {"id": "completion", "probe": "CompletionResolver",
-     "target": "bridged/src/main/java/dev/ltms/bridged/inject/CompletionResolver.java"},
+     "target": "fleetd/src/main/java/dev/ltms/fleet/inject/CompletionResolver.java"},
     {"id": "worker", "probe": "WorkerService",
-     "target": "bridged/src/main/java/dev/ltms/bridged/worker/WorkerService.java"},
+     "target": "fleetd/src/main/java/dev/ltms/fleet/worker/WorkerService.java"},
     {"id": "rendezvous", "probe": "Rendezvous",
-     "target": "bridged/src/main/java/dev/ltms/bridged/msg/Rendezvous.java"},
+     "target": "fleetd/src/main/java/dev/ltms/fleet/msg/Rendezvous.java"},
 ]
 
 PROMPT_TMPL = (
     "You are one of several issue-hunting workers in the claude-bridge repo (it is your "
     "current working directory). Your assignment: inspect the file `{target}` and find the "
     "SINGLE most important real bug, correctness gap, or risk in it. Read the file before "
-    "answering. Reply via bridge_reply with EXACTLY these four lines:\n"
+    "answering. Reply via fleet_reply with EXACTLY these four lines:\n"
     "1. {target}:<line>\n"
     "2. issue: <one sentence>\n"
     "3. fix: <one line>\n"
@@ -147,9 +147,9 @@ def grade(rec):
     phase, source, reply = rec["phase"], rec["source"], rec["reply"]
     has_reply = bool(reply and reply.strip())
     if phase == "done" and source == "reply" and has_reply:
-        return "OK", "clean explicit bridge_reply"
+        return "OK", "clean explicit fleet_reply"
     if phase == "done" and has_reply:
-        return "DEGRADED", f"resolved via {source} (worker did not call bridge_reply)"
+        return "DEGRADED", f"resolved via {source} (worker did not call fleet_reply)"
     if phase == "done" and not has_reply:
         return "EMPTY", "turn completed but reply was empty"
     if phase == "failed":

fleetd/.gitignore

@@ -0,0 +1,16 @@
+# Build output
+target/
+dependency-reduced-pom.xml
+
+# Local runtime config (copy from fleetd.example.yaml). Both names are ignored: fleetd.yaml is
+# the current name, and bridged.yaml is the legacy name Fleetd still falls back to.
+fleetd.yaml
+bridged.yaml
+
+# CB-505 audit trail + daemon stdout/stderr — runtime records, never source
+logs/
+
+# Editor / OS
+*.iml
+.idea/
+.DS_Store

fleetd/docs/CB-307-Push-Loop.md

@@ -5,10 +5,10 @@
 
 ## Why this exists
 
-Stage 2 gave a worker→primary reply a **durable place to wait** when no `bridge_send` is
+Stage 2 gave a worker→primary reply a **durable place to wait** when no `fleet_send` is
 open: it lands in `agent.<target>.inbox` on the broker and survives a daemon bounce. But
 delivery is still **pull** — the primary only sees the reply if it happens to call
-`bridge_poll(target)` / `GET /sessions/{id}/replies`. A reply can sit indefinitely while
+`fleet_poll(target)` / `GET /sessions/{id}/replies`. A reply can sit indefinitely while
 the primary works on something else.
 
 This layer makes delivery **active**: the bridge *pushes* a nudge to the primary the moment
@@ -26,11 +26,11 @@ pointed at the primary's pane instead.
 
 ```mermaid
 flowchart LR
-    W["worker"] -->|"bridge_reply (no open send)"| MS["MessageService.reply"]
+    W["worker"] -->|"fleet_reply (no open send)"| MS["MessageService.reply"]
     MS -->|"inbox.publish"| INBOX[("agent.&lt;target&gt;.inbox<br/>(durable, LavinMQ)")]
     MS -->|"notify"| LOOP["ReplyPushLoop"]
     LOOP -->|"status-gated inject"| PANE["primary's herdr pane"]
-    PANE -->|"primary drains"| DRAIN["bridge_poll(target)<br/>= peek + ack"]
+    PANE -->|"primary drains"| DRAIN["fleet_poll(target)<br/>= peek + ack"]
     DRAIN -->|"inbox now empty"| LOOP
     LOOP -.->|"still non-empty →<br/>re-inject on backoff"| PANE
     classDef store fill:#2c5282,stroke:#1a365d,color:#ffffff;
@@ -51,13 +51,13 @@ the caller runs in a herdr pane on this host. Today it's discarded for the prima
 (`presence.markPresent` is a no-op on it).
 
 **Plan:** a single-slot `PrimaryRegistry` (thread-safe) holding the primary's `terminal_id`.
-Populate it from the **orchestration-side** MCP tools — `bridge_send`, `bridge_spawn`,
-`bridge_poll`, `bridge_list`, `bridge_status`, `bridge_profiles` — capturing
+Populate it from the **orchestration-side** MCP tools — `fleet_send`, `fleet_spawn`,
+`fleet_poll`, `fleet_list`, `fleet_status`, `fleet_profiles` — capturing
 `callerTerminal(exchange)` when it is (a) non-null and (b) **not** a registered worker
 session in `SessionManager`. That caller is, by construction, the primary. Worker-side tools
-(`bridge_reply`, `bridge_ask`) never set it.
+(`fleet_reply`, `fleet_ask`) never set it.
 
-- **Config override / pin:** a `primary: { terminal: "<id>" }` block in `BridgedConfig`
+- **Config override / pin:** a `primary: { terminal: "<id>" }` block in `FleetConfig`
   (nested record, same shape as `Broker`). Lets an operator pin it, or supply it when
   derivation can't (see degrade case).
 - **Degrade:** if the primary is off-host or in a non-herdr terminal, `terminalForPid`
@@ -71,13 +71,13 @@ A `ReplyPushLoop` component, notified at the single no-waiter call site
 (`MessageService.reply` → the `inbox.publish` branch, `MessageService.java:192`).
 
 - **Inject a nudge, not the payload.** The injected turn tells the primary *to drain*
-  (e.g. "Worker `<target>` returned a reply — run `bridge_poll(target=<target>)` to collect
+  (e.g. "Worker `<target>` returned a reply — run `fleet_poll(target=<target>)` to collect
   it"), it does **not** carry the reply text. Rationale: replies can be large/multiline and
   terminal injection would mangle them; the drain response is the clean transport. Keeps the
   push idempotent — re-nudging is harmless.
 - **Ack = drain.** The primary draining (`drainReplies` = peek + ack) is the acknowledgement.
   The loop's **stop condition is `inbox.peek(target).isEmpty()`** — the reply is gone from the
-  inbox because it was acked. No new `bridge_ack` tool needed for v1 (see Increment 3).
+  inbox because it was acked. No new `fleet_ack` tool needed for v1 (see Increment 3).
 - **Status-gated injection (mechanism (b), chosen).** A dedicated lightweight scheduled loop,
   **not** the worker `Injector`. It injects via `AgentControl.send(primaryTerminal, nudge)`
   (the same herdr `agent.send` = `pane send-text` + submit that delivers to workers) only when
@@ -93,10 +93,10 @@ A `ReplyPushLoop` component, notified at the single no-waiter call site
   the reply remains in the durable inbox and the next natural poll (or a later worker reply's
   nudge) still surfaces it. Bounded so the bridge never spams the primary.
 
-### Increment 3 — optional per-`msgId` `bridge_ack` tool (deferred)
+### Increment 3 — optional per-`msgId` `fleet_ack` tool (deferred)
 
 Drain-as-ack is coarse: it clears *all* pending replies for a target at once. If finer
-control is ever needed (ack one reply, leave others held), add a `bridge_ack(msgId)` tool
+control is ever needed (ack one reply, leave others held), add a `fleet_ack(msgId)` tool
 mapping to `inbox.ack(target, msgId)` — the port already supports per-`msgId` ack. Not built
 in v1; the stop-on-empty loop is sufficient.
 
@@ -106,7 +106,7 @@ in v1; the stop-on-empty loop is sufficient.
    resolved terminal is non-null **and not a registered worker session**, seen on an
    orchestration-side tool. This never mislabels a worker (workers are in `SessionManager`)
    and needs no new env var or argument (identity stays connection-derived, per the existing
-   `BridgeMcp` invariant).
+   `FleetMcp` invariant).
 
 2. **Readiness-gate mismatch → dedicated loop.** The existing `Injector` gates delivery on
    `ready.test(target)` = `WorkerPresence` (the *worker's* MCP connected). The primary is not
@@ -132,7 +132,7 @@ boundary**. The bridge is signalling the primary that it has mail — not drivin
   non-null terminal AND not a registered session" predicate; the loop's stop-on-empty and
   bounded-reminder logic with an injected clock + a fake injector (no real herdr).
 - **Live dogfood (primary-side):** with the daemon on the broker jar + a real worker,
-  delegate a task, let the worker reply after the `bridge_send` window closes, and observe the
+  delegate a task, let the worker reply after the `fleet_send` window closes, and observe the
   bridge inject a drain nudge into *this* primary pane; confirm draining stops the reminders;
   confirm an unreachable primary (registry empty) degrades to pull with no loss.
 

fleetd/fleetd.example.yaml

@@ -1,10 +1,10 @@
-# bridged configuration (example). Copy to bridged.yaml and adjust.
+# fleetd configuration (example). Copy to fleetd.yaml and adjust.
 #
-# bridged is the sole gateway between primary/worker Claude sessions and herdr.
+# fleetd is the sole gateway between primary/worker Claude sessions and herdr.
 # It is NOT a Claude process and must never carry ANTHROPIC_BASE_URL.
 
 # REST + MCP listen address. Keep it on loopback unless you also switch auth.mode to `token`
-# below — bridged REFUSES TO START on a non-loopback bind under loopback-trust (see auth).
+# below — fleetd REFUSES TO START on a non-loopback bind under loopback-trust (see auth).
 bind:
   host: 127.0.0.1
   port: 8765
@@ -21,7 +21,7 @@ bind:
 #                           bind — the daemon fails fast otherwise, because "unauthenticated ⇒
 #                           primary" on a reachable port would hand spawn/stop/send to anyone.
 #   tokenEnv              → host env var holding the token (never the literal value). Default
-#                           BRIDGED_API_TOKEN. Read only in token mode; empty ⇒ startup fails.
+#                           FLEETD_API_TOKEN. Read only in token mode; empty ⇒ startup fails.
 #
 # TLS is deliberately NOT terminated in the daemon (CB-501 D3): run a reverse proxy in front and
 # let it own certificate lifecycle, e.g.
@@ -29,13 +29,13 @@ bind:
 # The broker link gets TLS from its own URI (amqps://…) — see `broker` below.
 # auth:
 #   mode: token
-#   tokenEnv: BRIDGED_API_TOKEN
+#   tokenEnv: FLEETD_API_TOKEN
 
 # Optional pinned primary terminal (CB-307). Names the herdr pane the PRIMARY itself runs in:
 # a caller whose connection maps to this pane resolves as the primary (no credential needed —
 # the pane mapping is as unforgeable as a worker's), and reply nudges are pushed to it.
 # REQUIRED when the primary runs inside a herdr pane — without it the pane match reads the
-# primary as a worker and refuses spawn/send/stop. Get the id from bridge_whoami; re-pin if
+# primary as a worker and refuses spawn/send/stop. Get the id from fleet_whoami; re-pin if
 # the primary moves panes.
 # primary:
 #   terminal: term_0123456789abcd
@@ -48,13 +48,13 @@ bind:
 # every orchestration call. List each lead's pane here and all of them resolve as leads.
 #
 #   tab → the ONLY field identity depends on (CB-579); the exact label of the tab hosting the lead.
-#         Label the tab yourself, or let bridged label one it launches — see `fleet.leaders:` below.
-#   kind/model → descriptive; they document what runs in the pane and are echoed by bridge_whoami
+#         Label the tab yourself, or let fleetd label one it launches — see `fleet.leaders:` below.
+#   kind/model → descriptive; they document what runs in the pane and are echoed by fleet_whoami
 #
-# A lead's tab must already carry its label (or be launched by bridged, which labels it) — there is
+# A lead's tab must already carry its label (or be launched by fleetd, which labels it) — there is
 # no terminal id to paste in and nothing to re-pin when the session restarts: the tab survives, so
 # the same label resolves the same lead again on the next scan.
-# `bridge_whoami` reports `{"role":"primary","leader":"<name>"}`; role stays "primary" because a lead
+# `fleet_whoami` reports `{"role":"primary","leader":"<name>"}`; role stays "primary" because a lead
 # IS a primary for authorization, so nothing that keys on the role breaks.
 #
 # KEEP `primary:` when adding leads: it still addresses the CB-307 push loop, which needs a single
@@ -63,13 +63,13 @@ bind:
 # Leads are configured under `fleet.leaders:` — see THE FLEET further down.
 #
 # Two things stop the tab-name convention from becoming a way to claim leadership: the configured
-# member spaces are excluded from the scan, so nothing bridged places can land in a matching tab;
+# member spaces are excluded from the scan, so nothing fleetd places can land in a matching tab;
 # and startup REFUSES a `tabPrefix` that the fleet tabLabel template, or any per-profile `tabLabel`
 # override, also matches — so the two namespaces cannot overlap by accident. The label is a NAME,
 # never a capability: what a pane may do is decided by the role the daemon resolves for it.
 
 # CB-551: IDLE-LEAD HEARTBEAT — nudge the single lead back to work when it has been continuously
-# idle (no open bridge_send driving it) past the quiet period. The fleet is one lead + architects +
+# idle (no open fleet_send driving it) past the quiet period. The fleet is one lead + architects +
 # workers, so a lead that stalls is a single point of failure; the ReplyPushLoop only nudges when a
 # reply lands, and this timer catches the gap where nothing lands and the lead just sits idle.
 #
@@ -93,21 +93,20 @@ bind:
 #   intervalSeconds → how often a tick runs (default 30). ENFORCED floor of 15: the code computes
 #                      Math.max(15, intervalSeconds), so a lower value is silently raised, not
 #                      rejected.
-#   workingSuspectAfterSeconds, paneProbeIntervalSeconds → accepted and parsed, but NOT YET READ by
-#                      anything — the dormant monitor only consumes intervalSeconds today (CB-573
-#                      shipped ahead of the evidence publishers these two knobs are for). Setting
-#                      them changes nothing right now, and no minimum is enforced on either, because
-#                      nothing reads them to enforce one. They exist so a later build can start
-#                      honouring them without another config-shape change.
-#   notifications.mode → "webhook" flips what bridge_list REPORTS (healthCoverage: "full" instead
-#                      of "detection-only") — it does NOT make bridged send any webhook call; no
+#   workingSuspectAfterSeconds → age before a BUSY member is suspected of a stall (default 600).
+#                      ENFORCED floor of 300: a lower value is silently raised.
+#   paneProbeIntervalSeconds → accepted and parsed, but NOT YET READ by anything. Setting it changes
+#                      nothing right now. It exists so a later build can start honouring it without
+#                      another config-shape change.
+#   notifications.mode → "webhook" flips what fleet_list REPORTS (healthCoverage: "full" instead
+#                      of "detection-only") — it does NOT make fleetd send any webhook call; no
 #                      delivery mechanism is implemented yet. Any other value, or omitting the
 #                      block, reports "detection-only".
 # health:
 #   enabled: true
-#   intervalSeconds: 30
-#   workingSuspectAfterSeconds: 600
-#   paneProbeIntervalSeconds: 60
+#   intervalSeconds: 30               # floor 15
+#   workingSuspectAfterSeconds: 600   # floor 300 — how long BUSY with no activity means STALL_SUSPECTED
+#   paneProbeIntervalSeconds: 60      # parsed, but nothing reads it yet — changing it changes nothing
 #   notifications:
 #     mode: disabled
 
@@ -115,6 +114,18 @@ bind:
 # (${HERDR_SOCKET_PATH:-~/.config/herdr/herdr.sock}).
 herdrSocket: ~/.config/herdr/herdr.sock
 
+# Optional socket for member panes. Omit this to use herdrSocket for both leads and members.
+# memberHerdrSocket: /Users/member/.config/herdr/herdr.sock
+
+# fleetd #213: the login shell the member OS user (memberHerdrSocket above) actually runs. ONLY
+# read when memberHerdrSocket is set — fleetd's own $SHELL says nothing about a pane running
+# under a different OS user, and there is no channel to ask herdr for that user's shell, so this
+# must be told rather than guessed. Absent, blank, or anything not ending in "zsh" is treated the
+# same as "not zsh": the memberCredentials.policy: allow-list ZDOTDIR scrub (see worktreeGroup
+# below) is skipped in favour of the weaker CB-596 sentinel overlay — a degraded control, never a
+# refusal to spawn. When memberHerdrSocket is absent this key is never consulted at all.
+# memberLoginShell: /bin/zsh
+
 # How member sessions are spawned. Define one or more named profiles (backends) under
 # `profiles`; each key is the profile name (also the ccs profile). A profile says only WHICH
 # BACKEND — model, CLI adapter, credentials, cost. It says nothing about what a member spawned on
@@ -124,8 +135,34 @@ herdrSocket: ~/.config/herdr/herdr.sock
 # Shared knobs (placement/workspace) can be repeated per profile; they usually match.
 #   placement: tab   → each worker lands in its OWN tab in a dedicated worker space (default).
 #                      Use `pane` for the legacy behaviour (split the focused tab).
-#   mcpUrl           → bridged mounts the bridge MCP (--mcp-config, inline) + reply charter
+#   mcpUrl           → fleetd mounts the bridge MCP (--mcp-config, inline) + reply charter
 #                      (--append-system-prompt) as launch flags; nothing is written to the profile.
+#   ideMcpUrl        → opt-in (CB-634), default off. When set, fleetd mounts the IDE Index MCP as a
+#                      second inline server named `intellij`, and adds an IDE charter that pins every
+#                      ide_* call to the member's own worktree. A URL, not a boolean — host and port
+#                      are host-specific. Set it only on a host where the IDE actually runs.
+#   ideProjectDir    → repo-relative module dir the IDE opens and the overlay pins (CB-634). Only read
+#                      when ideMcpUrl is set. This repo's Maven pom lives in `fleetd/`, not at the
+#                      worktree root, so opening the root imports no module and ide_* resolves nothing;
+#                      set this to `fleetd`. Omit for a repo whose project is the worktree root.
+#   ideOpenCommand   → host command that opens ideProjectDir in the IDE at spawn (CB-634 auto-open).
+#                      Only read when ideMcpUrl is set. `{dir}` is replaced with the absolute module
+#                      dir and the command runs through `/bin/sh -c`, so set env inline if needed —
+#                      e.g. `env DISPLAY=:10.0 idea {dir}`. Best-effort: a failure is logged, never
+#                      fails the spawn. Omit to open the member's module by hand. There is no close
+#                      half yet — an opened module stays open until the operator closes it.
+#   autoCompactWindow → opt-in, default off. A bounded token window that forces a spawned member to
+#                      compact its context instead of running on the backend's own default and dying
+#                      mid-turn (losing its fleet_reply — the whole point of the turn — with it).
+#                      Validated at config load to [100000, 1000000] — the band Claude Code's own
+#                      --autocompact flag accepts.
+#                      CROSS-BACKEND SEMANTICS DIFFER: on claude-code this is a launch-time
+#                      `--autocompact <tokens>` flag — the member compacts AT this window. opencode
+#                      has no equivalent flag (it only forces `compaction.auto: true`, unconditionally,
+#                      already), so this is instead applied as the model's `limit.context` in the
+#                      generated opencode.json — the member compacts WITHIN this window, not exactly
+#                      at it — and only when this profile's `model:` is in `provider/model` form; if it
+#                      isn't, fleetd logs a WARN naming the profile rather than silently doing nothing.
 #   tokenEnv         → host env var holding the worker's auth token (value never stored in config);
 #                      omit for a backend that needs no token (e.g. a local ollama).
 #   cwd              → pin this profile's working directory (CB-112). Omit to inherit the primary's
@@ -135,7 +172,12 @@ herdrSocket: ~/.config/herdr/herdr.sock
 #                      skills/MCP/hooks. Omit to leave the worker on the host default.
 #   parityOverlay    → repo-relative paths copied primary→worktree so a worker in a provisioned
 #                      worktree sees the same local config (CB-301-ext). Omit for the default set:
-#                      [.claude/settings.local.json, .env, .envrc].
+#                      [.env] only (CB-148). .envrc is left out of the default on purpose: it is
+#                      executable shell that direnv runs on every cd, so copying it carries
+#                      behaviour into the worker, not just values, unlike .env. An operator who
+#                      wants it copied can still write parityOverlay: [.env, .envrc] explicitly.
+#                      (.claude/settings.local.json is NOT in the default — it
+#                      pre-approves IDE/tool grants a member must not hold ambiently; CB-525/CB-634.)
 #
 #                      Do NOT add .mcp.json (CB-525). A worker's tools are whatever its launcher
 #                      mounts — the bridge, and nothing else. Replicating the primary's MCP config
@@ -143,7 +185,7 @@ herdrSocket: ~/.config/herdr/herdr.sock
 #                      checkout, so its navigation returned paths OUTSIDE its own worktree: one
 #                      worker made all 59 of its edits in the primary tree while compiling its
 #                      worktree, and every build it ran was of code that did not contain them.
-#                      bridged neutralizes a provisioned worktree's .mcp.json for this reason;
+#                      fleetd neutralizes a provisioned worktree's .mcp.json for this reason;
 #                      listing it here would copy the primary's back over that.
 #   gitTokenEnv      → host env var holding the git-forge API token. When set, its value is injected
 #                      as GITEA_TOKEN so the worker can open its OWN PR at checkpoint (CB-302).
@@ -152,11 +194,11 @@ herdrSocket: ~/.config/herdr/herdr.sock
 #   gitHostEnv       → host env var holding the forge host (default GITEA_HOST). Injected as
 #                      GITEA_HOST *only* alongside a resolved gitTokenEnv.
 #   exhaustedPattern → regex matched against a completion-fallback scrape (CB-578 stage A) to
-#                      classify a turn that ended with no bridge_reply as the backend having
+#                      classify a turn that ended with no fleet_reply as the backend having
 #                      refused on a subscription usage limit, rather than a real answer. Opt-in —
 #                      omit and this profile's completion fallback behaves exactly as before.
 #                      Every backend words its refusal differently, so this is config, never a
-#                      vendor string baked into bridged itself.
+#                      vendor string baked into fleetd itself.
 #                      DEFERRED: compiled once into a startup pattern map — editing it needs a
 #                      daemon restart, same as this profile's model/baseUrl/argv.
 #   credentialId     → CB-578 stage B: the credential this profile quarantines WITH when a
@@ -168,18 +210,51 @@ herdrSocket: ~/.config/herdr/herdr.sock
 #                      profile quarantines alone, under its own name, exactly as if the field did
 #                      not exist. Cooldown length is the top-level quarantineCooldownSeconds below.
 #                      HOT: read live at every spawn/exhaustion check — no restart needed.
+#   errorPattern     → fleetd #201 / #227: regex matched against a completion-fallback scrape to
+#                      classify a turn that ended with no fleet_reply as a BACKEND ERROR — a
+#                      credential outage or a provider 5xx — rather than a real answer or a
+#                      usage-limit exhaustion (exhaustedPattern above always wins when a line
+#                      matches both). Opt-in. Omit it and this profile falls back to fleetd's
+#                      built-in legacy pattern `(?i)\bAPI Error\s*:` — classification still
+#                      happens, just without a profile-specific match; every backend words its
+#                      failure differently, so a hardcoded sentence would only ever match one
+#                      of them.
+#                      DEFERRED: compiled once into a startup pattern map, same as exhaustedPattern
+#                      — editing it needs a daemon restart.
+#                      # errorPattern: "503 Service Unavailable"   # opt-in: classify a backend outage
+#
+#                      What happens once a match fires (BackendOutagePolicy, credentialId-keyed,
+#                      SEPARATE from the CB-578 stage B quarantine above and never merged with it):
+#                        - threshold 2 — TWO DISTINCT TARGETS (never raw events) on the same
+#                          effective credential inside a 60-second window start an "incident" and a
+#                          60-second cool-off for that credential. One member repeating the same
+#                          classified line twice never cools anything off — a real outage hits
+#                          every target on that credential, so requiring a second, independent
+#                          target loses nothing against the case this guards against, while
+#                          protecting against a heuristic misfire on one flaky member.
+#                        - a fresh error while a credential is already cooling off is ignored
+#                          outright: it neither extends the 60s deadline nor starts a new incident.
+#                        - `fleet_list`/`fleet_profiles` report a cooling credential with
+#                          `coolingOffForSeconds` (never `quarantinedForSeconds`, unless CB-578
+#                          exhaustion quarantine is ALSO independently active for the same
+#                          credential — the two checks can both fire at once). A spawn onto a
+#                          cooling profile is refused with a message naming the credential and
+#                          remaining seconds — "cooling off", never "exhausted", so an operator can
+#                          tell a short transient fault from a spent subscription at a glance.
+#                        - the lead gets ONE nudge per incident (not one per affected target), via
+#                          the same push loop that already delivers ticket/question reminders.
 #   env              → extra environment for this profile's workers, as a literal key/value map
 #                      (CB-511). Use it to give workers a toolchain.
 #
-#                      A worker's environment does NOT come from your shell. bridged hands herdr an
+#                      A worker's environment does NOT come from your shell. fleetd hands herdr an
 #                      explicit env map and herdr merges it into ITS OWN process env — so before
 #                      CB-511 a worker inherited whatever PATH the herdr server happened to be
 #                      started with, which on a long-lived herdr can predate your toolchain entirely
 #                      and leave workers unable to run `mvn` or `java` at all.
-#                      bridged now propagates ITS OWN PATH to every worker by default; set `env:`
+#                      fleetd now propagates ITS OWN PATH to every worker by default; set `env:`
 #                      only to override that or add more (JAVA_HOME, …). Since the default is the
 #                      daemon's PATH, make sure the daemon is started with a good one — see the PATH
-#                      lines in deploy/dev.ltms.bridged.plist and deploy/bridged.service.
+#                      lines in deploy/dev.ltms.fleet.plist and deploy/fleetd.service.
 #
 #                      Adapter-owned variables always win over `env:`: ANTHROPIC_BASE_URL and the
 #                      rest of the ANTHROPIC_*/CLAUDE_* wiring are applied after it, so an `env:`
@@ -192,20 +267,20 @@ profiles:
     baseUrl: http://gx01.gw:8000     # the vLLM host this profile targets (gx00.gw / gx01.gw)
     model: coder
     placement: tab
-    workspace: bridged-workers
+    workspace: fleetd-workers
     # tabLabel: an optional per-profile override; the fleet template usually covers it
     mcpUrl: http://127.0.0.1:8765/mcp
-    tokenEnv: BRIDGED_WORKER_TOKEN
+    tokenEnv: FLEETD_WORKER_TOKEN
     argv: ["ccs", "gx10"]
     # weight: relative selection weight for automatic placement (weighted, round-robin, and
     # fixed's fallback walk). Absent defaults to 1.0. An explicit 0 or negative value means
     # "never auto-select this profile" (CB-554) — it stays reachable via an explicit
-    # `bridge_spawn{profile:"gx10"}`, which bypasses placement entirely; only automatic
+    # `fleet_spawn{profile:"gx10"}`, which bypasses placement entirely; only automatic
     # selection skips it.
     weight: 0.5
     # maxLoad: max live workers on this profile. Omit for unlimited. An explicit 0 (CB-585) caps
     # the profile at zero live members — it is excluded from automatic placement and an explicit
-    # `bridge_spawn{profile:"gx10"}` against it is refused too; a cap holds even when the profile
+    # `fleet_spawn{profile:"gx10"}` against it is refused too; a cap holds even when the profile
     # is named directly. Negative is refused at config load — there is no sane meaning for it.
     maxLoad: 2
     # subscription: true
@@ -224,24 +299,58 @@ profiles:
     #   subscription path no guard would vet the URL, so allowing both would be a way around the
     #   guard rather than a configuration.
     #
-    #   GOTCHA 1 — it is invisible to the startup secret check. `Bridged.reportRequiredSecrets`
+    #   GOTCHA 1 — it is invisible to the startup secret check. `Fleetd.reportRequiredSecrets`
     #   skips subscription profiles on purpose (they need no token), so a boot log that reports
     #   every secret as fine says nothing about these profiles.
     #
     #   GOTCHA 2 — `maxLoad` is the ONLY throttle you have here. There is no metering, no budget
     #   and no refusal on cost; the cap on live members is the single thing standing between a
     #   fan-out and your monthly limit. Set it deliberately and keep it small.
+    #
+    #   GOTCHA 3 (fleetd #176) — `maxLoad` counts members, never the lead itself. The lead is a live
+    #   `claude` session on this SAME account (a lead is never moved off-subscription, whatever its
+    #   own profile says), so it already holds one seat before any member spawns. If a lead's
+    #   `fleet.leaders.<name>.profile` names THIS profile — or ANY OTHER `subscription: true`
+    #   profile that shares this one's account (see THE SENTINEL, just below, next to
+    #   `credentialId:`) — `fleet_list`'s `free` for this profile subtracts that lead's live
+    #   seat(s) automatically; see `profile:` under THE FLEET below. If no lead entry names a
+    #   profile sharing this account, fleetd has no way to know a lead holds a seat here, and `free`
+    #   will overstate what a fresh `fleet_spawn` actually gets by exactly the seats the lead is
+    #   quietly holding.
+    #
+    #   THE SENTINEL (fleetd #176 stage 2, correcting an inert stage 1 fix): every `subscription:
+    #   true` profile that leaves `credentialId` unset shares ONE implicit account-wide credential
+    #   id with every other such profile on this host — because a subscription profile doesn't
+    #   authenticate with a credential of its own, it authenticates as the operator's own Claude
+    #   login, and there is exactly one of those. So on a typical host, `opus` (the lead's profile)
+    #   and `sonnet` (the members' profile) are linked automatically, with NOTHING to set here — that
+    #   is what makes GOTCHA 3 above work without also writing matching `credentialId:` values on
+    #   both. This linkage is not just cosmetic: it is the same key `BackendQuarantine`/cool-off use,
+    #   so a usage-limit hit on `opus` now quarantines `sonnet` too (and vice versa) — correct, since
+    #   they are one Claude account, but worth knowing before you wonder why an unrelated-looking
+    #   profile went quarantined.
+    #
+    #   WHEN TO OVERRIDE — set explicit, DIFFERENT `credentialId:` values on two `subscription: true`
+    #   profiles only when they are genuinely two separate Claude logins on the same host (a real,
+    #   if unusual, setup). An explicit `credentialId` always wins over the sentinel, so this is the
+    #   one way to keep two subscription profiles from being treated as one account for lead-seat
+    #   counting AND for quarantine/cool-off grouping alike.
     # gitTokenEnv: GITEA_TOKEN        # opt-in: let this profile's workers open their own PR (CB-302)
     # gitHostEnv: GITEA_HOST          # defaults to GITEA_HOST; injected only with gitTokenEnv
     # exhaustedPattern: "usage limit has been reached"   # opt-in: classify a usage-limit refusal (CB-578)
     # credentialId: shared-openai   # opt-in: quarantine together with every other profile sharing this id (CB-578)
+    # errorPattern: "503 Service Unavailable"   # opt-in: classify a backend outage (fleetd #201/#227) — see the key doc above
     # configDir: /Users/me/.ccs/instances/gx10   # CLAUDE_CONFIG_DIR — inherit that profile's skills/MCP
     # cwd: /Users/me/src/myrepo       # pin the working dir; omit to inherit the primary's
-    # parityOverlay: [".claude/settings.local.json", ".env", ".envrc"]   # never add .mcp.json — see above
+    # parityOverlay: [".env"]   # the default; add ".envrc" explicitly if you want it copied too (CB-148) — never add .mcp.json or .claude/settings.local.json — see above
+    # ideMcpUrl: http://127.0.0.1:29170/index-mcp/streamable-http   # opt-in (CB-634): IDE code intelligence, pinned to the worktree
+    # ideProjectDir: fleetd          # CB-634: module dir the IDE opens + the overlay pins (this repo's pom is in fleetd/)
+    # ideOpenCommand: env DISPLAY=:10.0 idea {dir}   # CB-634 auto-open: opens {dir} in the IDE at spawn; omit to open by hand
+    # autoCompactWindow: 250000   # opt-in: bound member context; claude-code compacts AT this, opencode within it (model limit.context)
   gx11:                              # a second backend, so `placement: weighted` has a choice
     baseUrl: http://gx01.gw:8000     # self-hosted; ccs handles the model + token
     placement: tab
-    workspace: bridged-workers
+    workspace: fleetd-workers
     # tabLabel: an optional per-profile override; the fleet template usually covers it
     mcpUrl: http://127.0.0.1:8765/mcp
     argv: ["ccs", "gx11"]
@@ -259,15 +368,15 @@ profiles:
   # The bridge MCP + reply charter mount via a generated OPENCODE_CONFIG file, and the model is a
   # `provider/model` selector. Placement, tabs, cwd, and the readiness gate are shared with Claude.
   #
-  # Dogfood-verified 2026-07-29 against opencode 1.18.5 (spawn → readiness gate → bridge_send →
-  # structured bridge_reply → teardown). The `opencode/*-free` models run on opencode's own gateway
+  # Dogfood-verified 2026-07-29 against opencode 1.18.5 (spawn → readiness gate → fleet_send →
+  # structured fleet_reply → teardown). The `opencode/*-free` models run on opencode's own gateway
   # and need NO credentials — check `opencode models` for the current free list, since the names
   # change. That also makes the worker off-subscription by construction.
   #   opencode-free:
   #     kind: opencode
   #     model: opencode/north-mini-code-free  # `provider/model` selector, injected as `-m`
   #     placement: tab
-  #     workspace: bridged-workers
+  #     workspace: fleetd-workers
   #     tabLabel: "opencode: {profile} #{n}"
   #     mcpUrl: http://127.0.0.1:8765/mcp
   #     argv: ["opencode"]
@@ -292,7 +401,7 @@ profiles:
   #     baseUrl: http://127.0.0.1:8000
   #     model: local-vllm/deepseek-v4-flash
   #     placement: tab
-  #     workspace: bridged-workers
+  #     workspace: fleetd-workers
   #     tabLabel: "opencode: {profile} #{n}"
   #     mcpUrl: http://127.0.0.1:8765/mcp
   #     argv: ["opencode"]
@@ -327,11 +436,17 @@ placement: weighted
 # DEFERRED: baked once into the BackendQuarantine built at startup — a running quarantine keeps
 # its original cooldown regardless; a new value only applies to a quarantine that starts after a
 # restart. Editing this needs a daemon restart to take effect.
+#
+# This does NOT govern the fleetd #201 / #227 backend-error cool-off documented under errorPattern
+# above — that mechanism is a separate, shorter-lived, NOT-configurable policy (threshold 2 distinct
+# targets, 60-second window, 60-second cool-off), on purpose: it exists to survive a brief transient
+# fault, not to replace this 30-minute exhaustion quarantine. Do not conflate the two when reading
+# fleet_list/fleet_profiles — coolingOffForSeconds and quarantinedForSeconds are independent facts.
 # quarantineCooldownSeconds: 1800
 
 # Re-read this file without restarting the daemon (CB-559). Off unless you add this block, so an
-# upgraded bridged keeps the old behaviour: the file is read once at boot and never again.
-#   enabled          → turn the watch on. bridged checks the file's modified time on a timer and
+# upgraded fleetd keeps the old behaviour: the file is read once at boot and never again.
+#   enabled          → turn the watch on. fleetd checks the file's modified time on a timer and
 #                      reloads when it moves.
 #   intervalSeconds  → how often to check (default 10). One `stat` per tick, so this is cheap.
 #
@@ -342,7 +457,7 @@ placement: weighted
 #               / credentialId. Those are hot because the placement policy (and, for credentialId,
 #               the CB-578 stage B quarantine check) reads them through a supplier — being config is
 #               not by itself enough to make a key hot.
-#               EXCEPT `fleet.leaders`: Bridged.main reads it once at startup to build the lead tab
+#               EXCEPT `fleet.leaders`: Fleetd.main reads it once at startup to build the lead tab
 #               scanner and launcher, and neither is rebuilt on reload. A changed/added/removed
 #               `fleet.leaders` entry is silently accepted — the reload reports "config reloaded"
 #               with nothing in the deferred list — but has NO effect until you restart. Treat it
@@ -353,8 +468,10 @@ placement: weighted
 #               stage B — baked once into the quarantine tracker built at startup), ADDING or
 #               REMOVING a profile (a new backend needs its own launcher, and launchers are built
 #               once), AND an existing profile's launch settings — model, baseUrl, argv, env,
-#               configDir, mcpUrl, tabLabel, exhaustedPattern. The launcher takes a copy of
-#               `profiles:` at startup and resolves every spawn out of that copy, so those never
+#               configDir, mcpUrl, tabLabel, exhaustedPattern, errorPattern (fleetd #201 / #227 —
+#               compiled once into a startup pattern map the same way exhaustedPattern is). The
+#               launcher takes a copy of `profiles:` at startup and resolves every spawn out of
+#               that copy, so those never
 #               reach a launch until you restart. The reload logs them by name rather than
 #               pretending they applied.
 #   COLD      → cannot change at all: `bind:`, `herdrSocket:`, `broker:` and `auth:`. The socket is
@@ -415,6 +532,22 @@ fleet:
   # recognised: give it a `profile:` and the daemon launches the shortfall when fewer than
   # `instances` are live. Omit `profile:` and it is recognise-only, as before.
   #
+  # `profile:` has a SECOND job as of fleetd #176, even for a recognise-only lead you never want
+  # auto-launched: it is also how fleetd learns which account this lead's own session shares. A
+  # `subscription: true` profile bills the operator's Claude account, and the lead itself is always
+  # a live `claude` session on that same account — `maxLoad` never counted that seat. If a lead
+  # entry here names a profile that shares a worker profile's account, `fleet_list`'s `free` for
+  # that worker profile subtracts the lead's live seat(s) automatically. "Shares the account" is
+  # decided by matching `effectiveCredentialId()`, which (fleetd #176 stage 2 — see THE SENTINEL,
+  # next to `credentialId:`, in THE WORKERS above) means: an explicit, matching `credentialId:` on
+  # both, OR — the common case, needing NO extra config — both being `subscription: true` with
+  # `credentialId` left unset, since those all share one implicit account-wide id. A lead on `opus`
+  # and workers on `sonnet` link automatically this way; they do NOT need the same profile name.
+  # Setting `profile:` on an already-running, recognise-only lead is safe — the daemon only launches
+  # the SHORTFALL below `instances`, so naming a profile here does not, by itself, start anything.
+  # Omit it and fleetd has no way to derive the sharing — there is no other reliable signal on the
+  # daemon's side — so that lead's seat goes uncounted, exactly as before this ticket.
+  #
   # `tab:` (CB-579) is REQUIRED and is the only field identity depends on — the exact label of the
   # tab hosting the lead, matched case-insensitively. Label the tab yourself and put that same
   # string here, and the pane is recognised on the next rescan. Reopen the tab later, or the session
@@ -446,8 +579,8 @@ fleet:
   #     workspace: leads             # where a launched lead's tab is created (default "leads").
   #                                  # MUST NOT be a member workspace — those are excluded from the
   #                                  # scan, so a lead placed in one is never found again.
-  #     cwd: /path/to/repo           # the launched lead's working directory (default: bridged's own)
-  #     kind: claude                 # descriptive; reported by bridge_whoami
+  #     cwd: /path/to/repo           # the launched lead's working directory (default: fleetd's own)
+  #     kind: claude                 # descriptive; reported by fleet_whoami
   #   gpt-sol-5.6:
   #     tab: "lead: gpt-sol-5.6"
   #     kind: opencode
@@ -475,7 +608,7 @@ guard:
 
 # Member credential policy (CB-596, gitea issue #82). A herdr pane runs a LOGIN shell, and that
 # shell re-sources the operator's own secret store — so a spawned member inherits every credential
-# the operator's shell holds, not just the ones bridged means to give it. Measured on this host:
+# the operator's shell holds, not just the ones fleetd means to give it. Measured on this host:
 # 31 credential names, all set, with only ONE (GITEA_ACCESS_TOKEN) blocked before this — and that
 # block was a single name hardcoded in HerdrPeerLauncher.java, not driven by this file. This block
 # replaces that hardcoded shadow with a config-driven list of names.
@@ -505,20 +638,51 @@ guard:
 # through either: the daemon logs a WARN naming any credential-shaped env var it finds on neither
 # list (never its value), so a secret added to the store later does not go unnoticed forever.
 #
-#   policy → only "deny-by-default" exists today (an operator-authored deny-list was deliberately
-#            rejected — see above). An unrecognized value refuses to start, naming it.
-#   allow  → credential names a member legitimately needs. Left OUT of the pane's env overlay
-#            entirely, so the value the pane's own (login) shell exports passes through untouched.
-#   known  → every credential name the operator's store is known to export. Every name here NOT
-#            also in `allow` is overlaid with a non-secret sentinel value before the pane's login
-#            shell runs — real protection only for names the login shell does not itself re-export
-#            (see the ROUND-2 CORRECTION note above for the ones it does).
+#   policy → "deny-by-default" (the default; also accepted spelled "deny-list") overlays each
+#            known-but-not-allowed name BEFORE the pane's login shell runs — real protection only
+#            where that shell does not re-export the name (see ROUND-2 CORRECTION above). An
+#            unrecognized value refuses to start, naming it.
+#   policy → "allow-list" (CB-633) moves the control to a per-spawn ZDOTDIR directory the daemon
+#            generates and passes through tab.create's env map. Each generated startup file sources
+#            its ~/ counterpart FIRST and then runs the scrub, so the scrub happens after the
+#            operator's whole chain and no sourced file can undo it.
+#            The scrub is sourced from BOTH the generated .zshrc and the generated .zlogin, because
+#            herdr does not open the same kind of shell everywhere: macOS panes run a LOGIN zsh (so
+#            .zlogin runs), Linux panes run a plain interactive zsh (so .zlogin never runs at all).
+#            A scrub in .zlogin alone would be a control that silently does nothing on Linux.
+#            The allow-list is DERIVED, never typed:
+#            every profile's tokenEnv/gitTokenEnv/gitHostEnv values and env-map keys, plus an
+#            infrastructure set (PATH HOME SHELL TERM LANG LC_* TMPDIR USER LOGNAME PWD SHLVL EDITOR
+#            PAGER JAVA_HOME XDG_* ZDOTDIR), plus whatever keys this spawn's own env overlay carries.
+#            Adding a profile can therefore only widen the list, never break another spawn's scrub.
+#            Under this policy `known`/`allow` below become REPORTING ONLY — they feed the gap WARN,
+#            they are no longer a control. If the member's login shell is NOT zsh, the daemon logs a
+#            loud WARN saying protection is off and falls back to deny-by-default's overlay.
+#            Each pane writes a scrub-report.txt naming how many variables it kept of how many it
+#            saw; the daemon logs that "allowed N of M" line when the pane stops. If the report is
+#            MISSING the daemon logs a WARN instead — the scrub cannot then be confirmed to have
+#            run, and a silently-dead control is exactly what this policy exists to prevent.
+#   allow  → credential names a member legitimately needs. Under deny-by-default, left OUT of the
+#            pane's env overlay entirely, so the value the pane's own (login) shell exports passes
+#            through untouched. Under allow-list: reporting only.
+#   known  → every credential name the operator's store is known to export. Under deny-by-default,
+#            every name here NOT also in `allow` is overlaid with a non-secret sentinel value before
+#            the pane's login shell runs — real protection only for names that shell does not itself
+#            re-export (see the ROUND-2 CORRECTION note above). Under allow-list: reporting only.
+#   sshAuthSock → whether SSH_AUTH_SOCK may pass through under allow-list ("allow") or must be
+#            blanked like any other non-derived name ("block", the default). This is a decision you
+#            have to make explicitly: SSH_AUTH_SOCK is a handle to YOUR ssh-agent, and a member
+#            holding it can sign with your keys — it sits in no secret file and looks like no
+#            credential, which is why it slipped past three earlier tickets (gitea #110). Blocking
+#            it breaks git over SSH inside members (push/fetch authenticate as you); use HTTPS
+#            remotes or scoped deploy keys instead of allowing it lightly.
 #
 # HOT-RELOADABLE the same way `fleet:` is (CB-559): read fresh on every spawn, so editing this list
 # and reloading config (or restarting) changes what the NEXT spawn inherits; already-running members
 # are unaffected either way.
 # memberCredentials:
-#   policy: deny-by-default
+#   policy: deny-by-default        # or "deny-list", or "allow-list" (CB-633) — see above
+#   sshAuthSock: block             # allow-list only; see the sshAuthSock note above
 #   allow:
 #     - AI_GATEWAY_TOKEN     # named in a profile's tokenEnv (local/gx) — a member reaching the
 #                            # gateway is by design, not a leak
@@ -570,6 +734,24 @@ guard:
 # to a sibling directory of the repo root.
 # worktreeRoot: /Users/me/src/.bridged-worktrees
 
+# Worktree group sharing (fleetd #185 stage 3). OPTIONAL, off by default. Names an OS group
+# that a provisioned worktree's repo is made group-writable for (git config
+# core.sharedRepository group, plus a one-time chgrp/chmod/setgid fix-up), so a member spawned
+# under a DIFFERENT OS user (see memberHerdrSocket) can write its own worktree, its
+# per-worktree git metadata, and its own commit objects — without it, every file GitWorktrees
+# creates is owned by fleetd's own uid and unwritable by another user.
+# CAUTION: this isolates credentials, not the repository — a member in the group can still
+# write the operator's git objects and refs in the shared repo. The operator running fleetd
+# must already be a member of the named group, or every provisioning spawn fails loudly.
+#
+# fleetd #213: this is also the ONE group the memberCredentials.policy: allow-list ZDOTDIR scrub
+# reuses when memberHerdrSocket is set — deliberately not a second config key. Under
+# memberHerdrSocket, the scrub directory is generated under worktreeRoot (never java.io.tmpdir,
+# which the member OS user cannot reach) and shared read-only with this group. If worktreeGroup
+# is unset while memberHerdrSocket is set, the scrub cannot be guaranteed reachable by the member,
+# so fleetd falls back to the weaker CB-596 sentinel overlay instead (a WARN names the gap).
+# worktreeGroup: fleet-workers
+
 # Session lifecycle limits (CB-303). All knobs are opt-in; omit or set to null to keep
 # the feature disabled. By default the daemon never reaps, caps, or drains sessions.
 #   idleTtlSeconds     → reap READY/DONE sessions idle longer than this (never BUSY/SPAWNING)
@@ -593,14 +775,36 @@ guard:
 # RabbitMQ speaks the same AMQP 0-9-1, so it is a URI-only swap.
 #   uri      → AMQP connection URI. No trailing slash ⇒ the default vhost "/"; an empty path ("/")
 #              is vhost "" and will NOT connect. Encode a named vhost as .../%2Fmyvhost.
+#   uriEnv   → CB-151: name of a host env var holding the AMQP URI, preferred over `uri` (wins
+#              whenever set). The URI carries `user:pass@` inline, so naming a variable keeps the
+#              password out of fleetd.yaml — same pattern as auth.tokenEnv/Profile.tokenEnv. A
+#              uriEnv that resolves to an unset or blank variable is treated as NOT configured and
+#              the daemon falls back to the in-memory inbox, warning loudly.
 #   prefetch → CB-527: consumer basicQos, capping how many unacked messages the inbox holds
 #              in-heap per owned target (the rest sits on the broker's durable queue instead of
 #              growing the JVM heap). Default 32 when omitted.
 # broker:
-#   uri: amqp://guest:guest@127.0.0.1:5672
+#   uriEnv: LAVINMQ_URI
 #   prefetch: 32
 
-# Active push-to-primary (CB-307 Stage 3). When a worker reply lands with no open bridge_send,
+# Shared cross-host LEADER coordination broker. OMIT this block to leave lead-to-lead messaging
+# off entirely (config-only in this ticket — nothing here wires it into a live LeadMailbox yet).
+# This is a SEPARATE AMQP vhost from `broker:` above: member/worker inboxes always stay on the
+# per-fleet `broker:` vhost, and this vhost carries only leader-to-leader traffic, so two fleets
+# whose members must never see each other can still share one coordination vhost for their leads.
+#   uriEnv   → name of a host env var holding the coordination AMQP URI, same convention as
+#              broker.uriEnv (keeps the credential out of fleetd.yaml). Wins over `uri` when set.
+#   selfId   → this daemon's own lead coord-id — the name its mailbox is owned under
+#              (lead.<selfId>.inbox), e.g. "mac-opus" or "fleet01-lead". Must be globally unique
+#              across every daemon sharing this vhost.
+#   prefetch → consumer basicQos, capping how many unacked messages the mailbox holds in-heap.
+#              Default 32 when omitted.
+# coordinator:
+#   uriEnv: LEAD_COORD_URI
+#   selfId: mac-opus
+#   prefetch: 32
+
+# Active push-to-primary (CB-307 Stage 3). When a worker reply lands with no open fleet_send,
 # the ReplyPushLoop injects a *drain nudge* (never the payload) into the primary's own herdr
 # pane — status-gated (only when injectable, never mid-turn) and bounded. Ack = drain: the loop
 # stops as soon as the primary's inbox is empty.
@@ -611,11 +815,11 @@ guard:
 #
 #                    REQUIRED (CB-522) if the primary itself runs inside a herdr pane. Caller
 #                    identity resolves a loopback PID to its herdr pane, and PaneLocator scans
-#                    EVERY pane — not just bridged-spawned ones — so such a primary is otherwise
+#                    EVERY pane — not just fleetd-spawned ones — so such a primary is otherwise
 #                    classified as a WORKER and refused SPAWN/SEND/STOP. That failure is
 #                    self-locking: the learned terminal is populated by the very orchestration
 #                    calls being refused, so only this pinned value can break the cycle. Read the
-#                    id off bridge_whoami (it reports the current terminal even while
+#                    id off fleet_whoami (it reports the current terminal even while
 #                    misclassified) and re-pin whenever the primary moves panes.
 #   pushReminders  → max nudges before giving up (default 5)
 #   pushBackoffMs  → delay between nudges in ms (default 15000)

fleetd/pom.xml

@@ -5,17 +5,17 @@
     <modelVersion>4.0.0</modelVersion>
 
     <groupId>dev.ltms</groupId>
-    <artifactId>bridged</artifactId>
+    <artifactId>fleetd</artifactId>
     <version>1.0.0</version>
     <packaging>jar</packaging>
 
-    <name>bridged</name>
-    <description>claude-bridge message server: sole gateway between primary/worker Claude sessions and herdr</description>
+    <name>fleetd</name>
+    <description>fleet message server: sole gateway between a lead session, its members, and herdr</description>
 
     <properties>
         <maven.compiler.release>25</maven.compiler.release>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
-        <mainClass>dev.ltms.bridged.Bridged</mainClass>
+        <mainClass>dev.ltms.fleet.Fleetd</mainClass>
 
         <jackson.version>2.19.0</jackson.version>
         <javalin.version>6.7.0</javalin.version>
@@ -28,6 +28,7 @@
         <testcontainers.version>1.20.4</testcontainers.version>
         <commons-compress.version>1.27.1</commons-compress.version>
         <commons-lang3.version>3.18.0</commons-lang3.version>
+        <sqlite-jdbc.version>3.53.4.0</sqlite-jdbc.version>
     </properties>
 
     <!--
@@ -44,6 +45,12 @@
           3.0-rc5; bumping Jackson 3 to the patched 3.2.x breaks the SDK (annotation mismatch).
           Only the loopback /mcp endpoint parses this JSON, from trusted local Claude clients.
       The 11.0.23 -> 11.0.25 bump did clear jetty CVE-2024-8184 (5.9) and CVE-2024-6763.
+
+      fleetd #206: org.xerial:sqlite-jdbc 3.53.4.0 (added for OpenCodeSessionDiscovery) — the
+      only known advisory against this artifact is CVE-2023-32697 (RCE via an attacker-controlled
+      JDBC URL), fixed in 3.41.2.2; 3.53.4.0 is well past that fix and OSV.dev reports no open
+      advisory against it. Checked via the OSV.dev API (no Mend.io/JetBrains IDE MCP mount
+      available from this worktree) on 2026-08-31.
     -->
 
     <!-- Force the latest patched Jetty 11.x across all Javalin-pulled Jetty modules (no version
@@ -106,7 +113,7 @@
         </dependency>
 
         <!-- MCP server: the SERVER face. Streamable-HTTP servlet mounted on Javalin's Jetty at
-             /mcp, exposing bridge_send/bridge_reply/bridge_status as thin adapters over REST. -->
+             /mcp, exposing fleet_send/fleet_reply/fleet_status as thin adapters over REST. -->
         <dependency>
             <groupId>io.modelcontextprotocol.sdk</groupId>
             <artifactId>mcp</artifactId>
@@ -123,6 +130,17 @@
             <version>${amqp.version}</version>
         </dependency>
 
+        <!-- fleetd #206: opencode moved its session store from a JSON tree to SQLite
+             (opencode.db). This is the JDBC driver OpenCodeSessionDiscovery uses to read it
+             read-only. Ships bundled native libraries (linux/mac/windows, several archs), so it
+             is a heavier jar than most deps here — see the pom's dependency-security note below
+             for the size/CVE tradeoff actually measured. -->
+        <dependency>
+            <groupId>org.xerial</groupId>
+            <artifactId>sqlite-jdbc</artifactId>
+            <version>${sqlite-jdbc.version}</version>
+        </dependency>
+
         <!-- Logging -->
         <dependency>
             <groupId>org.slf4j</groupId>
@@ -161,7 +179,10 @@
     </dependencies>
 
     <build>
-        <finalName>bridged</finalName>
+        <!-- CB-634: the cutover renamed the module dir (bridged/ -> fleetd/), the jar, and the
+         launchd plist together. The installed plist names fleetd/target/fleetd.jar and
+         KeepAlive is armed, so this name, the plist, and the wrapper must move as one. -->
+        <finalName>fleetd</finalName>
         <plugins>
             <plugin>
                 <groupId>org.apache.maven.plugins</groupId>
@@ -203,7 +224,7 @@
                 </configuration>
             </plugin>
 
-            <!-- Runnable fat jar: java -jar target/bridged.jar -->
+            <!-- Runnable fat jar: java -jar target/fleetd.jar -->
             <plugin>
                 <groupId>org.apache.maven.plugins</groupId>
                 <artifactId>maven-shade-plugin</artifactId>

fleetd/src/main/java/dev/ltms/fleet/auth/AuditLog.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;

fleetd/src/main/java/dev/ltms/fleet/auth/Authz.java

@@ -1,9 +1,9 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
 /**
  * The authorization table (CB-505), stated once and enforced on both entry paths.
  *
- * <p>Most of these rules are already true de facto — {@code BridgeMcp} derives a worker's identity
+ * <p>Most of these rules are already true de facto — {@code FleetMcp} derives a worker's identity
  * from the connection rather than reading it from an argument, so a worker has never been able to
  * reply <em>as</em> another worker over MCP. What was missing is that the REST surface trusted the
  * session id in the URL path, and neither surface checked role at all. This class makes the

fleetd/src/main/java/dev/ltms/fleet/auth/CallerResolver.java

@@ -1,7 +1,7 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
-import dev.ltms.bridged.mcp.ConnectionIdentity;
-import dev.ltms.bridged.peer.MemberRole;
+import dev.ltms.fleet.mcp.ConnectionIdentity;
+import dev.ltms.fleet.peer.MemberRole;
 
 import java.nio.charset.StandardCharsets;
 import java.security.MessageDigest;
@@ -12,7 +12,7 @@ import java.util.function.Supplier;
 /**
  * Resolves every caller to a {@link Principal}, for both entry paths into the core (CB-501).
  *
- * <p>There are two of them and they are not layered the way the docs suggest: {@code BridgeMcp}
+ * <p>There are two of them and they are not layered the way the docs suggest: {@code FleetMcp}
  * calls the service layer directly and is mounted as a raw servlet (so it never passes through a
  * Javalin filter), while the REST routes historically resolved no identity at all. Both now
  * delegate here, so the authorization rules are stated once instead of drifting apart.
@@ -59,7 +59,7 @@ public final class CallerResolver {
      * <p>Like {@link #leadTerminals}, a supplier rather than a fixed map, so a binding injected
      * after startup — when the later spawn lifecycle establishes a live architect session, or an
      * operator pins one — takes effect without a restart. Consulted per resolve; today's wiring
-     * in {@code Bridged} reads a constant from config, which is the degenerate live case.
+     * in {@code Fleetd} reads a constant from config, which is the degenerate live case.
      */
     private final Supplier<Map<String, String>> architectTerminals;
     private final Function<String, MemberRole> memberSlotRoles;
@@ -235,7 +235,7 @@ public final class CallerResolver {
 
         // loopback-trust: same-host callers that are not workers are the primary. A non-loopback
         // caller is anonymous even here — and startup refuses that combination anyway
-        // (BridgedConfig.validateAuthExposure), so this is defence in depth, not the control.
+        // (FleetConfig.validateAuthExposure), so this is defence in depth, not the control.
         return isLoopback(remoteAddr) ? Principal.primary(c.pid()) : Principal.anonymous();
     }
 

fleetd/src/main/java/dev/ltms/fleet/auth/MemberLifecycle.java

@@ -0,0 +1,74 @@
+package dev.ltms.fleet.auth;
+
+import dev.ltms.fleet.peer.MemberRole;
+
+/** Optional session lifecycle hook for live member-slot bindings. */
+public interface MemberLifecycle {
+
+    /** A slot held before a member process starts. */
+    record SlotReservation(String slot, String profile) { }
+
+    MemberLifecycle NONE = new MemberLifecycle() {
+        @Override
+        public MemberRole acquired(MemberRole role, String profile, String terminal) {
+            return role; // no registry configured — nothing to bind against, so the request stands
+        }
+
+        @Override
+        public void released(String terminal) {
+        }
+
+        @Override
+        public void requireSlotFor(MemberRole role, String profile) {
+            // no registry configured — nothing to validate against, so nothing is refused
+        }
+
+        @Override
+        public SlotReservation reserve(MemberRole role, String profile) {
+            return null;
+        }
+
+        @Override
+        public boolean bind(SlotReservation reservation, String terminal) {
+            return false;
+        }
+
+        @Override
+        public void release(SlotReservation reservation) {
+        }
+    };
+
+    /**
+     * Try to bind a newly spawned {@code terminal} into the role it was granted.
+     *
+     * @return the role this session actually holds: {@code role} unchanged for a role with no
+     *         live slot-binding semantics (dev, reviewer), or when the bind succeeded; a fallback
+     *         role — never {@code role} — when a slot-bound role (architect) could not be bound.
+     *         Callers must record THIS value on the session, never the requested {@code role}, so
+     *         a later roster read never reports a role the session does not hold (CB-619). In
+     *         normal operation this fallback should not happen once a reservation has been bound.
+     *         It remains the honest answer if a caller has no reservation, or if binding a
+     *         reservation unexpectedly fails.
+     */
+    MemberRole acquired(MemberRole role, String profile, String terminal);
+
+    void released(String terminal);
+
+    /**
+     * Refuse an acquire before anything spawns when {@code role} requires a live slot binding and
+     * no configured slot carries {@code profile} (CB-619 / fleetd #123). A no-op for a role with
+     * no slot-binding semantics.
+     *
+     * @throws IllegalArgumentException naming the role, the profile, and the pools that do carry it
+     */
+    void requireSlotFor(MemberRole role, String profile);
+
+    /** Reserve a matching slot before launch, or refuse before a charter can be delivered. */
+    SlotReservation reserve(MemberRole role, String profile);
+
+    /** Convert a reservation into a live terminal binding. */
+    boolean bind(SlotReservation reservation, String terminal);
+
+    /** Return an unbound reservation after a failed launch. */
+    void release(SlotReservation reservation);
+}

fleetd/src/main/java/dev/ltms/fleet/auth/MemberRegistry.java

@@ -1,13 +1,14 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
-import dev.ltms.bridged.config.BridgedConfig;
-import dev.ltms.bridged.peer.MemberRole;
+import dev.ltms.fleet.config.FleetConfig;
+import dev.ltms.fleet.peer.MemberRole;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import java.util.Collections;
 import java.util.HashMap;
 import java.util.LinkedHashMap;
+import java.util.List;
 import java.util.Map;
 import java.util.Objects;
 
@@ -55,11 +56,13 @@ public final class MemberRegistry implements MemberLifecycle {
     }
 
     private final Map<String, Entry> slots;
-    /** Live {@code terminal_id → qualified slot key}; guarded by {@code this}. */
+    /** Live {@code terminal_id → qualified slot key}; guarded by {@code terminalToSlot}. */
     private final Map<String, String> terminalToSlot = new HashMap<>();
+    /** Slot keys held between reservation and the terminal binding. Guarded by terminalToSlot. */
+    private final java.util.Set<String> reservedSlots = new java.util.HashSet<>();
 
     /** Flatten every role pool in {@code fleet} into one registry. Leaders are not members. */
-    public MemberRegistry(BridgedConfig.Fleet fleet) {
+    public MemberRegistry(FleetConfig.Fleet fleet) {
         Map<String, Entry> flat = new LinkedHashMap<>();
         if (fleet != null) {
             for (MemberRole role : MemberRole.values()) {
@@ -94,7 +97,7 @@ public final class MemberRegistry implements MemberLifecycle {
      * An immutable copy of the live {@code terminal_id → slot name} bindings.
      *
      * <p>Passed to {@link CallerResolver} as the source of architect identity, and what
-     * {@code bridge_whoami}/the roster will read to say which slot a pane hosts. Empty until the
+     * {@code fleet_whoami}/the roster will read to say which slot a pane hosts. Empty until the
      * spawn lifecycle binds a slot.
      */
     public Map<String, String> snapshot() {
@@ -166,7 +169,7 @@ public final class MemberRegistry implements MemberLifecycle {
             if (existingSlot != null) {
                 return slot.equals(existingSlot); // already this slot (idempotent) or a different one
             }
-            if (terminalToSlot.containsValue(slot)) {
+            if (terminalToSlot.containsValue(slot) || reservedSlots.contains(slot)) {
                 return false; // slot already hosts a terminal — no second one
             }
             terminalToSlot.put(terminal, slot);
@@ -206,19 +209,116 @@ public final class MemberRegistry implements MemberLifecycle {
      *
      * <p>The role check is lifecycle policy. {@link CallerResolver} repeats it when resolving a
      * binding, so a later lifecycle regression cannot turn a worker into an architect.
+     *
+     * <p>CB-619 / fleetd #123: the return value is the role this session actually holds, and the
+     * caller is required to record THAT — never the requested {@code role} — on the session. Before
+     * this fix the caller kept the requested role regardless of whether the bind below succeeded, so
+     * a demoted session's {@code GET /members} row still said {@code "architect"} while
+     * {@code fleet_whoami} (which reads the live binding, not the request) correctly said
+     * {@code "worker"} — three sources of truth that disagreed about one live member, silently.
      */
     @Override
-    public void acquired(MemberRole role, String profile, String terminal) {
+    public MemberRole acquired(MemberRole role, String profile, String terminal) {
         if (role != MemberRole.ARCHITECT || terminal == null || terminal.isBlank()) {
-            return;
+            return role;
         }
         // slotsFor preserves definition order, so duplicate-profile slots use the first free one.
         for (Entry entry : slotsFor(MemberRole.ARCHITECT).values()) {
             if (Objects.equals(profile, entry.profile()) && bind(entry.key(), terminal)) {
-                return;
+                return MemberRole.ARCHITECT;
+            }
+        }
+        // fleetd #123: at least WARN — a role downgrade that the roster must now also reflect is
+        // not routine bookkeeping. requireSlotFor already refuses the config-gap case (no slot at
+        // all carries this profile) before a process ever spawns; reaching here means the config DID
+        // carry a matching slot but every one of them was already bound to a different terminal — a
+        // race this pre-spawn check cannot close on its own (see requireSlotFor's javadoc).
+        log.warn("member slot: no free architect slot for profile={} terminal={}; holding the session "
+                + "as {} instead of the architect it asked for — every configured slot for this "
+                + "profile is already bound to a different terminal", profile, terminal,
+                MemberRole.DEV.wireName());
+        return MemberRole.DEV;
+    }
+
+    /**
+     * CB-619 / fleetd #123: refuse an architect acquire before anything spawns when no configured
+     * slot carries {@code profile} — the config-gap case from the original defect report (a spawn
+     * asked for {@code role=architect, profile=sonnet}, and {@code fleet.architects} carried only
+     * {@code opus} and {@code sol}). A dev/reviewer acquire is always a no-op: those pools are
+     * placement candidates only (see {@code CompositePeerLauncher}), never a live identity binding,
+     * so there is nothing here to refuse — an explicit profile outside the pool for those roles is a
+     * documented operator override, not a defect.
+     *
+     * <p>This closes the config-gap case, not the live-capacity case: a profile that DOES carry a
+     * slot can still lose the race to a concurrent spawn between this check and the actual
+     * {@link #bind}, which is why {@link #acquired} must still answer honestly even after this
+     * check has passed.
+     */
+    @Override
+    public void requireSlotFor(MemberRole role, String profile) {
+        if (role != MemberRole.ARCHITECT) {
+            return;
+        }
+        boolean hasSlot = slotsFor(MemberRole.ARCHITECT).values().stream()
+                .anyMatch(e -> Objects.equals(profile, e.profile()));
+        if (hasSlot) {
+            return;
+        }
+        List<String> pools = slotsFor(MemberRole.ARCHITECT).values().stream()
+                .map(Entry::profile)
+                .distinct()
+                .toList();
+        throw new IllegalArgumentException(
+                "no " + role.wireName() + " slot for profile '" + profile + "' — an architect's "
+                        + "identity IS the slot it is bound to, so there is nothing to bind this "
+                        + "session's identity to. fleet." + role.configKey() + " carries profiles: "
+                        + (pools.isEmpty() ? "(none configured)" : String.join(", ", pools))
+                        + "; add profile '" + profile + "' there, or spawn " + role.wireName()
+                        + " on one of those profiles instead");
+    }
+
+    @Override
+    public SlotReservation reserve(MemberRole role, String profile) {
+        if (role != MemberRole.ARCHITECT) {
+            return null;
+        }
+        synchronized (terminalToSlot) {
+            for (Entry entry : slotsFor(MemberRole.ARCHITECT).values()) {
+                if ((profile == null || profile.isBlank() || Objects.equals(profile, entry.profile()))
+                        && !terminalToSlot.containsValue(entry.key()) && reservedSlots.add(entry.key())) {
+                    return new SlotReservation(entry.key(), entry.profile());
+                }
+            }
+        }
+        throw new IllegalArgumentException("no free architect slot for profile '" + profile
+                + "' — every matching slot is already bound or reserved");
+    }
+
+    @Override
+    public boolean bind(SlotReservation reservation, String terminal) {
+        if (reservation == null || terminal == null || terminal.isBlank()) {
+            return false;
+        }
+        synchronized (terminalToSlot) {
+            if (!reservedSlots.remove(reservation.slot())) {
+                return false;
+            }
+            if (!isSlot(reservation.slot()) || terminalToSlot.containsKey(terminal)
+                    || terminalToSlot.containsValue(reservation.slot())) {
+                return false;
+            }
+            terminalToSlot.put(terminal, reservation.slot());
+            return true;
+        }
+    }
+
+    @Override
+    public void release(SlotReservation reservation) {
+        if (reservation != null) {
+            synchronized (terminalToSlot) {
+                reservedSlots.remove(reservation.slot());
             }
         }
-        log.info("member slot: no free architect slot for profile={}; session remains a worker", profile);
     }
 
     /** Unbind a released terminal using the compare-safe registry operation. */

fleetd/src/main/java/dev/ltms/fleet/auth/Principal.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
 /**
  * A resolved caller: its {@link Role}, and — for a worker — the herdr {@code terminal_id} that
@@ -39,14 +39,14 @@ public record Principal(Role role, String terminal, long pid, String name) {
      *
      * <p>Carries {@link Role#PRIMARY}: a lead <em>is</em> a primary as far as authorization goes,
      * so every existing {@code isPrimary()} gate keeps working unchanged and the role table needed
-     * no new entry. The name is reporting only — it lets {@code bridge_whoami} say <em>which</em>
+     * no new entry. The name is reporting only — it lets {@code fleet_whoami} say <em>which</em>
      * lead is asking once more than one is configured.
      *
      * <p><strong>CB-532: a lead now carries the terminal it was matched by.</strong> Under CB-530 it
      * deliberately did not, because {@code terminal} meant "which worker pane" everywhere and a
      * non-null one would have enrolled the lead in the worker presence map. That reading was what
      * made a lead unaddressable: {@link #ownsSession} could never be true for it, so
-     * {@code bridge_reply} was refused and one lead could send to another but never be answered.
+     * {@code fleet_reply} was refused and one lead could send to another but never be answered.
      * The terminal now means "which pane is this caller", the presence map keys on
      * {@link #isSpawnedMember()} instead, and a lead is a peer that can both send and receive.
      */
@@ -63,7 +63,7 @@ public record Principal(Role role, String terminal, long pid, String name) {
      * An architect (CB-548), identified by the slot it occupies and the pane bound to it.
      *
      * <p>Carries {@link Role#ARCHITECT}. {@code slotName} is reporting only — it lets
-     * {@code bridge_whoami} say <em>which</em> architect slot is asking, and it is the key the
+     * {@code fleet_whoami} say <em>which</em> architect slot is asking, and it is the key the
      * (future) spawn lifecycle reads a profile back from. Identity is the {@code terminal}: like a
      * worker's it comes from the connection and the live terminal→slot binding, so
      * {@code ownsSession} works exactly as it does for a worker — an architect acts as its own

fleetd/src/main/java/dev/ltms/fleet/auth/Role.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.auth;
+package dev.ltms.fleet.auth;
 
 /**
  * What a caller is allowed to be on the bus (CB-501).

fleetd/src/main/java/dev/ltms/fleet/config/ConfigRef.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.config;
+package dev.ltms.fleet.config;
 
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -16,7 +16,7 @@ import java.util.function.Supplier;
 /**
  * The daemon's live configuration, re-readable without a restart (CB-559).
  *
- * <p>Consumers hold this, not a {@link BridgedConfig}, and read through {@link #get()} at the point
+ * <p>Consumers hold this, not a {@link FleetConfig}, and read through {@link #get()} at the point
  * of use. A component that captures {@code ref.get()} into a field at construction has opted out of
  * reload — which is sometimes right (see <em>deferred</em> below), but it must then be a deliberate
  * choice rather than an accident of where the field was initialised.
@@ -31,7 +31,7 @@ import java.util.function.Supplier;
  *       {@code placement:}, and an existing profile's {@code weight} / {@code maxLoad}. Those
  *       three are read through a supplier on {@code CompositePeerLauncher}, which is what makes
  *       them hot — not the fact that they are config. <strong>This does NOT include
- *       {@code fleet.leaders}</strong>: {@code Bridged.main} reads {@code cfg.fleet().leaders()}
+ *       {@code fleet.leaders}</strong>: {@code Fleetd.main} reads {@code cfg.fleet().leaders()}
  *       once at startup to build the {@code LeadTabScanner} and the {@code LeadLauncher}, and
  *       neither is reconstructed on reload — so a lead added, removed, or re-{@code tab}'d under
  *       {@code fleet.leaders} needs a restart, the same as any deferred key below.</li>
@@ -42,8 +42,10 @@ import java.util.function.Supplier;
  *       {@code guard:}, {@code worktreeRoot:}, adding or removing a profile (a new backend needs its own launcher,
  *       which is constructed once), <em>and an existing profile's launch settings</em> —
  *       {@code model}, {@code baseUrl}, {@code argv}, {@code env}, {@code mcpUrl},
- *       {@code exhaustedPattern} (CB-578 stage A — compiled once into {@code Bridged.main}'s
- *       pattern map at startup), and the rest. {@code credentialId} (CB-578 stage B) is NOT on
+ *       {@code exhaustedPattern} (CB-578 stage A — compiled once into {@code Fleetd.main}'s
+ *       pattern map at startup), {@code errorPattern} (fleetd #201 Unit 5 — compiled once into
+ *       {@code Fleetd.main}'s backend-error pattern map at startup, the same way), and the rest.
+ *       {@code credentialId} (CB-578 stage B) is NOT on
  *       this list — it is read live off the config supplier at every quarantine check and
  *       exhaustion event, exactly like {@code weight} / {@code maxLoad}, so it is hot instead.
  *       {@code HerdrPeerLauncher} takes {@code Map.copyOf(profiles)} at construction and resolves
@@ -65,30 +67,30 @@ import java.util.function.Supplier;
  * running. A config file being edited is normally read once mid-save; degrading a working daemon
  * because it caught a half-written file would be a bad trade.
  */
-public final class ConfigRef implements Supplier<BridgedConfig> {
+public final class ConfigRef implements Supplier<FleetConfig> {
 
     private static final Logger log = LoggerFactory.getLogger(ConfigRef.class);
 
     /** Keys that cannot change under a running daemon — see the class doc. */
     private static final Set<String> COLD_KEYS =
-            Set.of("bind", "herdrSocket", "broker", "auth");
+             Set.of("bind", "herdrSocket", "memberHerdrSocket", "broker", "auth");
 
     private final Path path;
-    private final AtomicReference<BridgedConfig> current;
+    private final AtomicReference<FleetConfig> current;
 
-    public ConfigRef(Path path, BridgedConfig initial) {
+    public ConfigRef(Path path, FleetConfig initial) {
         this.path = path;
         this.current = new AtomicReference<>(Objects.requireNonNull(initial, "initial config"));
     }
 
     /** A fixed reference that never reloads — for tests and for wiring built from a config in code. */
-    public static ConfigRef fixed(BridgedConfig cfg) {
+    public static ConfigRef fixed(FleetConfig cfg) {
         return new ConfigRef(null, cfg);
     }
 
     /** The live configuration. Read this per use; do not cache it in a field. */
     @Override
-    public BridgedConfig get() {
+    public FleetConfig get() {
         return current.get();
     }
 
@@ -128,7 +130,7 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
             }
             if (!applied) {
                 return "config reload refused — these keys cannot change under a running daemon: "
-                        + String.join(", ", coldKeys) + ". Restart bridged to apply them.";
+                        + String.join(", ", coldKeys) + ". Restart fleetd to apply them.";
             }
             if (!deferred.isEmpty()) {
                 return "config reloaded; these changes need a restart to take effect: "
@@ -149,10 +151,10 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
         if (path == null) {
             return Outcome.failed("this config was built in code and has no file to reload from");
         }
-        BridgedConfig old = current.get();
-        BridgedConfig fresh;
+        FleetConfig old = current.get();
+        FleetConfig fresh;
         try {
-            fresh = BridgedConfig.load(path);
+            fresh = FleetConfig.load(path);
             // The same gate startup runs. A config that would have refused to boot must not be able
             // to slip in through a reload — that is how a daemon ends up in a state it could never
             // have started in, which is the hardest kind to debug.
@@ -182,7 +184,7 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
     }
 
     /** Cold keys whose value differs between the running config and the candidate. */
-    private static List<String> changedColdKeys(BridgedConfig old, BridgedConfig fresh) {
+    private static List<String> changedColdKeys(FleetConfig old, FleetConfig fresh) {
         List<String> changed = new ArrayList<>();
         if (!Objects.equals(old.bind(), fresh.bind())) {
             changed.add("bind");
@@ -190,6 +192,9 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
         if (!Objects.equals(old.herdrSocket(), fresh.herdrSocket())) {
             changed.add("herdrSocket");
         }
+        if (!Objects.equals(old.memberHerdrSocket(), fresh.memberHerdrSocket())) {
+            changed.add("memberHerdrSocket");
+        }
         if (!Objects.equals(old.broker(), fresh.broker())) {
             changed.add("broker");
         }
@@ -202,7 +207,7 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
     }
 
     /** Changed keys that were accepted but whose effect waits for a restart. */
-    private static List<String> changedDeferredKeys(BridgedConfig old, BridgedConfig fresh) {
+    private static List<String> changedDeferredKeys(FleetConfig old, FleetConfig fresh) {
         List<String> changed = new ArrayList<>();
         if (!Objects.equals(old.lifecycle(), fresh.lifecycle())) {
             changed.add("lifecycle");
@@ -226,9 +231,9 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
         if (!Objects.equals(old.quarantineCooldownSeconds(), fresh.quarantineCooldownSeconds())) {
             changed.add("quarantineCooldownSeconds");
         }
-        Map<String, BridgedConfig.Profile> before =
+        Map<String, FleetConfig.Profile> before =
                 old.profiles() == null ? Map.of() : old.profiles();
-        Map<String, BridgedConfig.Profile> after =
+        Map<String, FleetConfig.Profile> after =
                 fresh.profiles() == null ? Map.of() : fresh.profiles();
         // Adding or removing a profile is deferred: a new backend needs its own launcher, and
         // launchers are built once at startup.
@@ -247,7 +252,7 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
         // a reload can produce, because the operator has no reason to doubt it.
         List<String> relaunch = new ArrayList<>();
         before.forEach((name, was) -> {
-            BridgedConfig.Profile now = after.get(name);
+            FleetConfig.Profile now = after.get(name);
             if (now != null && !sameLaunchSettings(was, now)) {
                 relaunch.add(name);
             }
@@ -266,7 +271,7 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
      * {@code CompositePeerLauncher}/the CB-578 stage B exhaustion sink) and really do take effect on
      * the next spawn.
      */
-    private static boolean sameLaunchSettings(BridgedConfig.Profile a, BridgedConfig.Profile b) {
+    private static boolean sameLaunchSettings(FleetConfig.Profile a, FleetConfig.Profile b) {
         return Objects.equals(a.baseUrl(), b.baseUrl())
                 && Objects.equals(a.model(), b.model())
                 && Objects.equals(a.configDir(), b.configDir())
@@ -276,6 +281,9 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
                 && Objects.equals(a.workspace(), b.workspace())
                 && Objects.equals(a.tabLabel(), b.tabLabel())
                 && Objects.equals(a.mcpUrl(), b.mcpUrl())
+                // CB-634: the IDE MCP mount is a launch flag, fixed at spawn like mcpUrl — a
+                // reload changes it only for members spawned after, so a changed value is deferred.
+                && Objects.equals(a.ideMcpUrl(), b.ideMcpUrl())
                 && Objects.equals(a.cwd(), b.cwd())
                 && Objects.equals(a.parityOverlay(), b.parityOverlay())
                 && Objects.equals(a.gitTokenEnv(), b.gitTokenEnv())
@@ -283,9 +291,13 @@ public final class ConfigRef implements Supplier<BridgedConfig> {
                 && Objects.equals(a.kind(), b.kind())
                 && Objects.equals(a.env(), b.env())
                 && Objects.equals(a.subscription(), b.subscription())
-                // CB-578 stage B: exhaustedPattern is compiled once into Bridged.main's pattern map
+                // CB-578 stage B: exhaustedPattern is compiled once into Fleetd.main's pattern map
                 // at startup (see ExhaustedPatternLookup wiring) — a reload never re-reads it, so a
                 // changed pattern must be reported as deferred, exactly like model/baseUrl/argv.
-                && Objects.equals(a.exhaustedPattern(), b.exhaustedPattern());
+                && Objects.equals(a.exhaustedPattern(), b.exhaustedPattern())
+                // fleetd #201 Unit 5: errorPattern is compiled once into Fleetd.main's backend-error
+                // pattern map at startup (see BackendErrorPatternLookup wiring), the same way
+                // exhaustedPattern is — a reload never re-reads it either.
+                && Objects.equals(a.errorPattern(), b.errorPattern());
     }
 }

fleetd/src/main/java/dev/ltms/fleet/config/ConfigWatcher.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.config;
+package dev.ltms.fleet.config;
 
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -11,7 +11,7 @@ import java.util.concurrent.ScheduledExecutorService;
 import java.util.concurrent.TimeUnit;
 
 /**
- * Polls {@code bridged.yaml}'s modified time and asks {@link ConfigRef} to reload when it moves
+ * Polls {@code fleetd.yaml}'s modified time and asks {@link ConfigRef} to reload when it moves
  * (CB-559). Opt-in through {@code configReload.enabled}.
  *
  * <p><strong>Why polling and not a filesystem watch.</strong> {@code WatchService} on macOS has no

fleetd/src/main/java/dev/ltms/fleet/guard/GuardException.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.guard;
+package dev.ltms.fleet.guard;
 
 /** Thrown when the subscription boundary would be violated. Never swallow this. */
 public class GuardException extends RuntimeException {

fleetd/src/main/java/dev/ltms/fleet/guard/SubscriptionGuard.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.guard;
+package dev.ltms.fleet.guard;
 
 import java.net.URI;
 import java.net.URISyntaxException;
@@ -16,7 +16,7 @@ import java.util.Set;
  *       means its traffic would leave the subscription. That is a hard stop.</li>
  * </ul>
  *
- * Both checks throw {@link GuardException} on violation. {@code bridged} calls
+ * Both checks throw {@link GuardException} on violation. {@code fleetd} calls
  * {@link #assertWorker} before spawning a worker and {@link #assertPrimaryClean}
  * against its own environment at startup.
  */

fleetd/src/main/java/dev/ltms/fleet/health/FleetHealth.java

@@ -1,7 +1,7 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
-import dev.ltms.bridged.herdr.AgentStatus;
-import dev.ltms.bridged.session.MemberSession;
+import dev.ltms.fleet.herdr.AgentStatus;
+import dev.ltms.fleet.session.MemberSession;
 
 /**
  * Pure classifier. Collection and repair are deliberately outside this package.

fleetd/src/main/java/dev/ltms/fleet/health/FleetHealthMonitor.java

@@ -1,10 +1,11 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
-import dev.ltms.bridged.herdr.Agent;
-import dev.ltms.bridged.herdr.AgentControl;
-import dev.ltms.bridged.herdr.AgentStatus;
-import dev.ltms.bridged.msg.MessageService;
-import dev.ltms.bridged.session.MemberSession;
+import dev.ltms.fleet.herdr.Agent;
+import dev.ltms.fleet.herdr.AgentControl;
+import dev.ltms.fleet.herdr.AgentStatus;
+import dev.ltms.fleet.herdr.HerdrException;
+import dev.ltms.fleet.msg.MessageService;
+import dev.ltms.fleet.session.MemberSession;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -25,6 +26,8 @@ public final class FleetHealthMonitor {
 
     /** Bounded attempts to run {@link #failTarget} for one transition. Never retried tick-to-tick (CB-580). */
     static final int MAX_FAIL_TARGET_ATTEMPTS = 3;
+    // CB-641: Match the injector's 60s readiness gate so health allows a full first boot.
+    static final long READINESS_GRACE_NANOS = TimeUnit.SECONDS.toNanos(60);
 
     private final AgentControl agents;
     private final Supplier<List<MemberSession>> roster;
@@ -32,29 +35,48 @@ public final class FleetHealthMonitor {
     private final ScheduledExecutorService scheduler;
     private final LongSupplier clock;
     private final long intervalSeconds;
+    private final long workingSuspectAfterNanos;
     private final BiConsumer<String, String> failTarget;
     private final Map<String, HealthPrior> priors = new HashMap<>();
     private final Map<String, HealthState> states = new HashMap<>();
+    /**
+     * CB-643: consecutive ticks on which a target looked like an orphaned delegation. The fact
+     * {@link MessageService#hasOrphanedDelegation} reports is a true snapshot, but it can read true
+     * for one tick during an ordinary race — an async ticket exists before its virtual thread has
+     * reached {@code rendezvous.open()}, so for that instant nothing is accepted or queued behind
+     * it. {@code decide} maps the field straight to {@code DELEGATION_ORPHANED} with no cross-tick
+     * smoothing of its own, so a single racy read would log a fault that clears on the next tick.
+     * Requiring two consecutive observations costs one interval of latency on a real orphan and
+     * removes that false positive entirely.
+     */
+    private final Map<String, Integer> orphanStreaks = new HashMap<>();
 
-    // These facts need the evidence publishers introduced by later M4 units. They are not negatives.
-    private static final boolean NOT_YET_OBSERVED = false;
+    /** How many consecutive ticks a target must look orphaned before health reports it (CB-643). */
+    static final int ORPHAN_CONFIRM_TICKS = 2;
+
+    // CB-643: every HealthSnapshot field now carries real evidence. The NOT_YET_OBSERVED placeholder
+    // that stood in for 7 of the 12 is gone, and with it the reason 8 of the 9 fault states were
+    // unreachable — GONE and NEVER_READY included, which is what kept CB-580's failTarget from ever
+    // firing. Do not reintroduce a constant here: a field with no publisher is a dead state, and the
+    // tests pass either way, so nothing else will tell you.
 
     /**
      * @param failTarget CB-568's idempotent target-wide failure operation (e.g. {@code messages::abandon}),
      *                   invoked once when a member transitions into a terminal health state. Required —
      *                   there is deliberately no defaulting overload; a caller that does not want the
      *                   fail-tickets-on-terminal-health behavior must pass an explicit inert value (see
-     *                   {@code TestTurnTokens.inert} / {@code BridgeMcp.CapacitySource.none()} for the pattern).
+     *                   {@code TestTurnTokens.inert} / {@code FleetMcp.CapacitySource.none()} for the pattern).
      */
     public FleetHealthMonitor(AgentControl agents, Supplier<List<MemberSession>> roster, MessageService messages,
                               ScheduledExecutorService scheduler, LongSupplier clock, long intervalSeconds,
-                              BiConsumer<String, String> failTarget) {
+                              long workingSuspectAfterSeconds, BiConsumer<String, String> failTarget) {
         this.agents = agents;
         this.roster = roster;
         this.messages = messages;
         this.scheduler = scheduler;
         this.clock = clock;
         this.intervalSeconds = intervalSeconds;
+        this.workingSuspectAfterNanos = TimeUnit.SECONDS.toNanos(workingSuspectAfterSeconds);
         this.failTarget = Objects.requireNonNull(failTarget, "failTarget");
     }
 
@@ -69,28 +91,50 @@ public final class FleetHealthMonitor {
     // Package-private so tests can run one tick without waiting.
     void tick() {
         try {
-            List<Agent> agentsNow = agents.list(); // Exactly one list call for this complete observation.
             List<MemberSession> rosterNow = roster.get(); // One in-memory roster snapshot for this tick.
+            List<Agent> agentsNow;
+            boolean controlLinkDown = false;
+            try {
+                agentsNow = agents.list(); // Exactly one list call for this complete observation.
+            } catch (HerdrException error) {
+                agentsNow = List.of();
+                controlLinkDown = true;
+                log.warn("fleet health control link unavailable; classifying roster", error);
+            }
             Map<String, Agent> live = new HashMap<>();
             for (Agent agent : agentsNow) live.put(agent.terminalId(), agent);
             HashSet<String> current = new HashSet<>();
+            long nowNanos = clock.getAsLong();
             for (MemberSession session : rosterNow) {
                 current.add(session.terminalId());
                 Agent agent = live.get(session.terminalId());
                 AgentStatus status = agent == null ? AgentStatus.UNKNOWN : agent.status();
                 boolean accepted = messages.hasAcceptedDelivery(session.terminalId());
-                HealthSnapshot snapshot = new HealthSnapshot(session.state(), status, accepted, NOT_YET_OBSERVED,
-                        messages.hasInboxMessage(session.terminalId()), agent != null, NOT_YET_OBSERVED,
-                        NOT_YET_OBSERVED, NOT_YET_OBSERVED, NOT_YET_OBSERVED, NOT_YET_OBSERVED, NOT_YET_OBSERVED);
+                boolean present = agent != null;
+                boolean targetNotFound = !controlLinkDown && !present
+                        && session.state() != MemberSession.State.SPAWNING;
+                boolean readinessGraceElapsed = nowNanos - session.spawnedAtNanos() >= READINESS_GRACE_NANOS;
+                boolean stalled = session.state() == MemberSession.State.BUSY
+                        && nowNanos - session.lastActivityAtNanos() >= workingSuspectAfterNanos;
+                // CB-643: the three message-layer facts CB-640 published. Read them here rather than
+                // leaving them false — that constant is what made 8 of the 9 fault states dead.
+                boolean queuedDelivery = messages.hasQueuedDelivery(session.terminalId());
+                boolean replyStranded = messages.hasStrandedReply(session.terminalId());
+                boolean orphanedDelegation = confirmOrphan(session.terminalId(),
+                        messages.hasOrphanedDelegation(session.terminalId()));
+                HealthSnapshot snapshot = new HealthSnapshot(session.state(), status, accepted, queuedDelivery,
+                        messages.hasInboxMessage(session.terminalId()), present, targetNotFound, controlLinkDown,
+                        readinessGraceElapsed, orphanedDelegation, replyStranded, stalled);
                 HealthDecision decision = decide(snapshot, priors.getOrDefault(session.terminalId(), HealthPrior.NONE),
-                        clock.getAsLong());
+                        nowNanos);
                 priors.put(session.terminalId(), decision.prior());
                 reportTransition(session.terminalId(), decision.state());
             }
             priors.keySet().retainAll(current);
             states.keySet().retainAll(current);
+            orphanStreaks.keySet().retainAll(current);
         } catch (Throwable error) {
-            // A list failure is health evidence, and must never kill the monitor's only scheduler task.
+            // Any unclassified collection failure must never kill the monitor's only scheduler task.
             log.warn("fleet health collection failed; will retry next tick", error);
         } finally {
             if (!scheduler.isShutdown()) {
@@ -99,6 +143,20 @@ public final class FleetHealthMonitor {
         }
     }
 
+    /**
+     * Debounce {@link MessageService#hasOrphanedDelegation} across ticks (CB-643). Returns true only
+     * once {@code observed} has held for {@link #ORPHAN_CONFIRM_TICKS} consecutive ticks; a single
+     * false reading resets the streak, so a transient race never reaches the classifier.
+     */
+    private boolean confirmOrphan(String target, boolean observed) {
+        if (!observed) {
+            orphanStreaks.remove(target);
+            return false;
+        }
+        int streak = orphanStreaks.merge(target, 1, Integer::sum);
+        return streak >= ORPHAN_CONFIRM_TICKS;
+    }
+
     void reportTransition(String target, HealthState next) {
         HealthState previous = states.put(target, next);
         if (previous == next) return;

fleetd/src/main/java/dev/ltms/fleet/health/HealthDecision.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
 /** Classification plus the private fact that the next pure decision needs. */
 public record HealthDecision(HealthState state, HealthPrior prior) { }

fleetd/src/main/java/dev/ltms/fleet/health/HealthPrior.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
 /** Private cross-tick observation. It is deliberately not a reported health value. */
 public record HealthPrior(boolean busyButDone) {

fleetd/src/main/java/dev/ltms/fleet/health/HealthSnapshot.java

@@ -1,7 +1,7 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
-import dev.ltms.bridged.herdr.AgentStatus;
-import dev.ltms.bridged.session.MemberSession;
+import dev.ltms.fleet.herdr.AgentStatus;
+import dev.ltms.fleet.session.MemberSession;
 
 /** Read-only facts from one fleet collection tick. */
 public record HealthSnapshot(MemberSession.State sessionState, AgentStatus liveStatus,

fleetd/src/main/java/dev/ltms/fleet/health/HealthState.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
 /** Health classifications reported for a member. */
 public enum HealthState {

fleetd/src/main/java/dev/ltms/fleet/health/MuteCounter.java

@@ -1,12 +1,12 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
-import dev.ltms.bridged.msg.Rendezvous;
+import dev.ltms.fleet.msg.Rendezvous;
 
 import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
 
 /**
- * Counts turns that ended via the completion fallback instead of {@code bridge_reply}.
+ * Counts turns that ended via the completion fallback instead of {@code fleet_reply}.
  * MUTE is an observation by target and profile, not a classifier state and never suppresses faults.
  */
 public final class MuteCounter {

fleetd/src/main/java/dev/ltms/fleet/health/PaneBudget.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.health;
+package dev.ltms.fleet.health;
 
 import java.util.ArrayList;
 import java.util.HashMap;

fleetd/src/main/java/dev/ltms/fleet/herdr/Agent.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 

fleetd/src/main/java/dev/ltms/fleet/herdr/AgentControl.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 
@@ -38,6 +38,11 @@ public final class AgentControl {
         this.herdr = herdr;
     }
 
+    /** The herdr daemon this control object sends its agent calls to. */
+    public HerdrClient herdr() {
+        return herdr;
+    }
+
     /** One agent-targeted call, translating a terminal id to its pane id (retrying once fresh). */
     private JsonNode agentCall(String method, String target, Map<String, Object> extra) {
         String resolved = resolveTarget(target);

fleetd/src/main/java/dev/ltms/fleet/herdr/AgentStatus.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 /**
  * A herdr agent's lifecycle state, as reported by {@code agent_status}. Drives the
@@ -32,7 +32,7 @@ public enum AgentStatus {
         };
     }
 
-    /** Whether {@code bridged} may inject a message now without stepping on a live turn. */
+    /** Whether {@code fleetd} may inject a message now without stepping on a live turn. */
     public boolean injectable() {
         return this == IDLE || this == BLOCKED || this == DONE;
     }

fleetd/src/main/java/dev/ltms/fleet/herdr/HerdrClient.java

@@ -1,17 +1,17 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 
 /**
  * Client face onto the herdr daemon (protocol 14, herdr 0.7.0).
  *
- * <p>This is the ONLY thing in {@code bridged} that speaks to herdr. Every method
+ * <p>This is the ONLY thing in {@code fleetd} that speaks to herdr. Every method
  * maps to a herdr JSON-RPC call over its Unix domain socket. Requests are
  * newline-delimited JSON with a <em>string</em> id; responses carry either a
  * {@code result} object (whose {@code type} field discriminates the payload) or an
  * {@code error} object.
  *
- * <p>Higher layers ({@code bridged}'s policy brain, REST endpoints, MCP adapters)
+ * <p>Higher layers ({@code fleetd}'s policy brain, REST endpoints, MCP adapters)
  * depend on this interface, not on the socket. Tests substitute a fake; the
  * {@code contract}-tagged suite exercises the real implementation against a running
  * herdr to catch protocol drift.

fleetd/src/main/java/dev/ltms/fleet/herdr/HerdrCodec.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.core.JsonProcessingException;
 import com.fasterxml.jackson.databind.JsonNode;

fleetd/src/main/java/dev/ltms/fleet/herdr/HerdrException.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 /**
  * Raised when a herdr call fails: transport error, or an {@code error} envelope

fleetd/src/main/java/dev/ltms/fleet/herdr/HerdrRouter.java

@@ -0,0 +1,40 @@
+package dev.ltms.fleet.herdr;
+
+import java.util.Objects;
+import java.util.function.Predicate;
+
+/** Routes lead operations and member operations to their owning herdr daemon. */
+public final class HerdrRouter implements AutoCloseable {
+    private final HerdrClient lead;
+    private final HerdrClient member;
+    private final AgentControl leadAgents;
+    private final AgentControl memberAgents;
+    private final WorkspaceControl leadSpaces;
+    private final WorkspaceControl memberSpaces;
+    private final Predicate<String> isLead;
+
+    public HerdrRouter(HerdrClient lead, HerdrClient member, Predicate<String> isLead) {
+        this.lead = Objects.requireNonNull(lead, "lead");
+        this.member = member != null ? member : lead;
+        this.isLead = Objects.requireNonNull(isLead, "isLead");
+        leadAgents = new AgentControl(this.lead);
+        memberAgents = this.member == this.lead ? leadAgents : new AgentControl(this.member);
+        leadSpaces = new WorkspaceControl(this.lead);
+        memberSpaces = this.member == this.lead ? leadSpaces : new WorkspaceControl(this.member);
+    }
+
+    public AgentControl leadAgents() { return leadAgents; }
+    public WorkspaceControl leadSpaces() { return leadSpaces; }
+    public AgentControl memberAgents() { return memberAgents; }
+    public WorkspaceControl memberSpaces() { return memberSpaces; }
+    public AgentControl agentsFor(String targetId) { return isLead.test(targetId) ? leadAgents : memberAgents; }
+
+    HerdrClient leadClient() { return lead; }
+    HerdrClient memberClient() { return member; }
+
+    @Override
+    public void close() {
+        lead.close();
+        if (member != lead) member.close();
+    }
+}

fleetd/src/main/java/dev/ltms/fleet/herdr/LeadTabScanner.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 import org.slf4j.Logger;
@@ -14,7 +14,7 @@ import java.util.function.Supplier;
 
 /**
  * Discovers which panes host a lead by scanning herdr for tabs the operator labelled by convention
- * (CB-531), and hands {@link dev.ltms.bridged.auth.CallerResolver} the resulting
+ * (CB-531), and hands {@link dev.ltms.fleet.auth.CallerResolver} the resulting
  * {@code terminal_id → lead name} map.
  *
  * <p><strong>Why scan at all.</strong> A lead is never spawned — a human opens a tab and starts an
@@ -39,18 +39,18 @@ import java.util.function.Supplier;
  *   <li>Worker spaces are excluded wholesale ({@code excludedWorkspaceLabels}), so a worker cannot
  *       become a lead by being placed — as a split, say — inside a matching tab.</li>
  *   <li>A worker cannot rename a tab: {@code tab.rename} is reachable only through
- *       {@link WorkspaceControl}, which no {@code bridge_*} tool exposes. The label is writable by
+ *       {@link WorkspaceControl}, which no {@code fleet_*} tool exposes. The label is writable by
  *       the human at the terminal and by nobody the bridge is defending against.</li>
  *   <li>The label is a <em>name</em>, not a capability. What a pane may do is decided by
  *       {@code Authz} against the role {@code CallerResolver} returns; a tab that calls itself a
  *       lead still cannot act as one unless the daemon's own registry agrees.</li>
  * </ol>
  *
- * <p><strong>CB-558 — bridged now writes lead labels too.</strong> This class used to be able to say
- * that bridged never renames a lead tab, so the label was always the human's own writing and there
+ * <p><strong>CB-558 — fleetd now writes lead labels too.</strong> This class used to be able to say
+ * that fleetd never renames a lead tab, so the label was always the human's own writing and there
  * was no round-trip from the daemon's rename back into its next decision.
- * {@code dev.ltms.bridged.lead.LeadLauncher} ends that: an auto-launched lead is labelled by the
- * daemon and found again by this scan. The trust direction above is unaffected — bridged writing a
+ * {@code dev.ltms.fleet.lead.LeadLauncher} ends that: an auto-launched lead is labelled by the
+ * daemon and found again by this scan. The trust direction above is unaffected — fleetd writing a
  * name for a lead it just started is not a pane promoting itself — but <em>staleness</em> becomes
  * real: a label left behind by a session that has since died would read as a live lead forever.
  * This scanner does not solve that (its job is naming, and a stale name costs nothing here); the
@@ -58,7 +58,7 @@ import java.util.function.Supplier;
  * ever make a decision that <em>removes</em> something based on this map, add the same check.
  * The remaining hazard is an <em>operator</em> one — a worker {@code tabLabel} template that
  * happens to start with the same prefix would promote the whole fleet — and that is refused at
- * startup by {@code BridgedConfig.validateLeadTabPrefixes} rather than documented here.
+ * startup by {@code FleetConfig.validateLeadTabPrefixes} rather than documented here.
  *
  * <p><strong>Caching.</strong> {@link #get()} is on the request path (every resolve), so the scan
  * is TTL-cached and a stale-but-valid map is preferred to a herdr round-trip. A failed scan keeps

fleetd/src/main/java/dev/ltms/fleet/herdr/PaneLocator.java

@@ -0,0 +1,147 @@
+package dev.ltms.fleet.herdr;
+
+import com.fasterxml.jackson.databind.JsonNode;
+
+import java.util.LinkedHashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.OptionalLong;
+import java.util.Set;
+
+/**
+ * Resolves which herdr pane a process belongs to — the herdr half of connection-based MCP
+ * identity (CB-105). Given the PID that opened an MCP connection, {@link #terminalForPid} finds
+ * the agent pane whose process tree contains it, so {@code fleetd} can tell <em>which worker</em>
+ * is calling without the worker sending anything spoofable.
+ *
+ * <p>herdr owns the PID→pane truth: {@code pane.process_info} reports each pane's {@code shell_pid}
+ * and foreground process PIDs. A pid that is neither of those directly — e.g. a grandchild a
+ * worker spawned, such as a {@code python3} or {@code curl} helper that opens its own MCP
+ * connection — is resolved by walking its ancestry (via {@link ParentResolver}) up to the root and
+ * matching any ancestor against a pane's {@code shell_pid} or foreground pids (CB-161). Without
+ * this walk such a pid matches no pane, and the caller falls through to loopback-trust and is
+ * resolved as the primary — a worker→primary privilege escalation.
+ *
+ * <p>This scans agent panes; a spawn-time {@code pid→terminal} cache is the obvious optimization
+ * once wired into {@code ClaudeCodeLauncher}.
+ *
+ * <p>CB-185 split the fleet across two herdr daemons — lead operations on one, members on the
+ * other ({@code memberHerdrSocket}). A caller's pane can live on <em>either</em> daemon (a lead's
+ * MCP connection resolves against the lead daemon; a member's against the member daemon), so this
+ * must be able to search more than one client. {@link #PaneLocator(HerdrClient, HerdrClient)}
+ * searches the lead client first, then the member client, and collapses to a single scan when the
+ * two are the same object (the historical single-daemon deployment). The caller's ancestor set is
+ * computed once per {@link #terminalForPid} call and reused across every client searched — it
+ * does not depend on which daemon a pane happens to live on.
+ */
+public final class PaneLocator {
+
+    /**
+     * Bound on how many ancestor generations {@link #ancestorsOf} walks. This runs on every MCP
+     * call, so a cycle or a pathologically deep process tree must not hang identity resolution;
+     * 32 generations is far more than any real worker→helper process tree needs.
+     */
+    private static final int MAX_ANCESTRY_DEPTH = 32;
+
+    private final List<HerdrClient> herdrs;
+    private final ParentResolver parentResolver;
+
+    /** Search only this client — the single-daemon deployment. */
+    public PaneLocator(HerdrClient herdr) {
+        this(herdr, ParentResolver.PROCESS_HANDLE);
+    }
+
+    /** Search only this client, resolving ancestry through {@code parentResolver} — for tests. */
+    public PaneLocator(HerdrClient herdr, ParentResolver parentResolver) {
+        this.herdrs = List.of(herdr);
+        this.parentResolver = parentResolver;
+    }
+
+    /**
+     * Search {@code lead} first, then {@code member} — the two-daemon deployment (CB-185). When
+     * the caller passes the same client for both (no {@code memberHerdrSocket} configured), this
+     * collapses to one client and one scan, exactly {@link #PaneLocator(HerdrClient)}'s behaviour.
+     */
+    public PaneLocator(HerdrClient lead, HerdrClient member) {
+        this(lead, member, ParentResolver.PROCESS_HANDLE);
+    }
+
+    /** Two-daemon deployment, resolving ancestry through {@code parentResolver} — for tests. */
+    public PaneLocator(HerdrClient lead, HerdrClient member, ParentResolver parentResolver) {
+        this.herdrs = lead == member ? List.of(lead) : List.of(lead, member);
+        this.parentResolver = parentResolver;
+    }
+
+    /**
+     * The {@code terminal_id} of the agent pane whose process tree contains {@code pid}, or
+     * {@code null} if no agent pane on any searched daemon owns it (e.g. the caller is the
+     * primary, or off-host).
+     */
+    public String terminalForPid(long pid) {
+        if (pid <= 0) {
+            return null;
+        }
+        Set<Long> ancestry = ancestorsOf(pid);
+        for (HerdrClient herdr : herdrs) {
+            String terminal = terminalForPid(herdr, ancestry);
+            if (terminal != null) {
+                return terminal;
+            }
+        }
+        return null;
+    }
+
+    /**
+     * {@code pid} itself plus its ancestor chain, walked through {@link #parentResolver} up to
+     * {@link #MAX_ANCESTRY_DEPTH} generations or pid 1, whichever comes first. A vanished ancestor
+     * ({@link ParentResolver#parentOf} returning empty) ends the walk without error — it just means
+     * the chain is shorter than the bound. A cycle in a fake resolver is caught by the "already
+     * seen" check and also ends the walk, so this can never loop.
+     */
+    private Set<Long> ancestorsOf(long pid) {
+        Set<Long> ancestry = new LinkedHashSet<>();
+        long current = pid;
+        for (int depth = 0; depth < MAX_ANCESTRY_DEPTH; depth++) {
+            if (current <= 0 || !ancestry.add(current)) {
+                break; // vanished/invalid pid, or a cycle back to a pid already recorded
+            }
+            if (current == 1) {
+                break; // reached the root of the process tree
+            }
+            OptionalLong parent = parentResolver.parentOf(current);
+            if (parent.isEmpty()) {
+                break; // vanished ancestor — not an error, just the end of the chain
+            }
+            current = parent.getAsLong();
+        }
+        return ancestry;
+    }
+
+    private static String terminalForPid(HerdrClient herdr, Set<Long> ancestry) {
+        for (JsonNode pane : herdr.call("pane.list", Map.of()).path("panes")) {
+            String paneId = pane.path("pane_id").asText(null);
+            if (paneId != null && paneOwnsAnyOf(herdr, paneId, ancestry)) {
+                return pane.path("terminal_id").asText(null);
+            }
+        }
+        return null;
+    }
+
+    private static boolean paneOwnsAnyOf(HerdrClient herdr, String paneId, Set<Long> ancestry) {
+        JsonNode info;
+        try {
+            info = herdr.call("pane.process_info", Map.of("pane_id", paneId)).path("process_info");
+        } catch (HerdrException e) {
+            return false; // pane vanished mid-scan — just skip it
+        }
+        if (ancestry.contains(info.path("shell_pid").asLong(-1))) {
+            return true;
+        }
+        for (JsonNode p : info.path("foreground_processes")) {
+            if (ancestry.contains(p.path("pid").asLong(-1))) {
+                return true;
+            }
+        }
+        return false;
+    }
+}

fleetd/src/main/java/dev/ltms/fleet/herdr/ParentResolver.java

@@ -0,0 +1,21 @@
+package dev.ltms.fleet.herdr;
+
+import java.util.OptionalLong;
+
+/**
+ * Resolves a pid's parent pid — the seam {@link PaneLocator} walks a process's ancestry through,
+ * so its tests can drive the walk from a fake pid→parent map instead of spawning real processes.
+ *
+ * <p>{@link #PROCESS_HANDLE} is the production implementation, backed by {@link ProcessHandle}.
+ */
+public interface ParentResolver {
+
+    /** The parent pid of {@code pid}, or empty if {@code pid} is gone or has no known parent. */
+    OptionalLong parentOf(long pid);
+
+    /** Production resolver: asks the JVM's {@link ProcessHandle} view of the OS process tree. */
+    ParentResolver PROCESS_HANDLE = pid -> ProcessHandle.of(pid)
+            .flatMap(ProcessHandle::parent)
+            .map(parent -> OptionalLong.of(parent.pid()))
+            .orElse(OptionalLong.empty());
+}

fleetd/src/main/java/dev/ltms/fleet/herdr/Tab.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 

fleetd/src/main/java/dev/ltms/fleet/herdr/UnixSocketHerdrClient.java

@@ -1,4 +1,4 @@
-package dev.ltms.bridged.herdr;
+package dev.ltms.fleet.herdr;
 
 import com.fasterxml.jackson.databind.JsonNode;
 import com.fasterxml.jackson.databind.ObjectMapper;