CB-622: rename the opencode mount key bridged -> fleetd

Unit C could not make this change. The tracked opencode.json is neutralized
by the worktree overlay (GitWorktrees.WORKTREE_HOSTILE_CONFIGS), so every
worker sees a stub instead of the real file and correctly reported that it
held no mount key. The lead has to make this one by hand, the same way
bridged.yaml is handled.

Takes effect for an opencode lead on its next session start.

.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 bridged.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 `bridge_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 bridged.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 `bridge_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 bridged.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 `bridge_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.

.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 bridged.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 `bridge_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 bridged.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 `bridge_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 bridged.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 `bridge_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

@@ -78,7 +78,7 @@ below are the procedure — run them in order, every task, not only the big ones
    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{ticket, msgId}`. Answer a worker's `bridge_ask`
+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`
    that answers it. **A worker's ask waits ~55 seconds, and no nudge makes that longer** — so never
@@ -173,12 +173,14 @@ you.
 |---|---|---|
 | 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 |
 | **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)
 
@@ -192,8 +194,10 @@ charter, not here.
 - **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, kept out of this file because it loads into every
-  session's context.
+  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.
 
 ### Redeploying the daemon — the lead may do this (primary only)
 

README.md

@@ -18,8 +18,9 @@ one unified Claude setup and the **sole communication gateway** (REST/SSE stays
 clients; any broker is `bridged`-internal, below the gateway).
 herdr owns the PTYs, multiplexing, persistence, and **agent-status events**; `bridged` owns
 policy (subscription boundary, session lifecycle, status-gated delivery) and the client
-contract. The worker `claude` launches with `ANTHROPIC_BASE_URL=https://ollama.ltms.dev` + a
-bearer token; the primary Opus stays env-clean and calls `bridged`'s MCP tools.
+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.
 
 ```mermaid
 flowchart LR
@@ -31,7 +32,7 @@ flowchart LR
     end
     HERDR["herdr<br/>panes · agent-status"]
     W["worker claude pane<br/>ANTHROPIC_BASE_URL set<br/>MCP client"]
-    M["ollama.ltms.dev<br/>(worker model)"]
+    M["llm.ltms.dev<br/>(the one gateway)"]
 
     OPUS -->|"MCP bridge_send (blocks)"| SRV
     W -.->|"MCP bridge_reply"| SRV

bridged/bridged.example.yaml

@@ -208,6 +208,29 @@ profiles:
     # `bridge_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
+    #   THE KNOB THAT DECIDES WHO PAYS (CB-539). Default false. When true, this profile's members
+    #   run on the OPERATOR'S OWN Claude subscription instead of a metered endpoint — every spawn
+    #   bills your plan and eats your usage limit. Off-subscription is the whole point of this
+    #   daemon, so treat `true` as a deliberate exception, not a convenience.
+    #
+    #   What changes when it is set (ClaudeCodeLauncher):
+    #     - no ANTHROPIC_BASE_URL and no ANTHROPIC_AUTH_TOKEN are injected — the member inherits
+    #       the operator's own Claude Code auth, which is exactly why it bills the plan;
+    #     - SubscriptionGuard never vets it, because there is no baseUrl to vet;
+    #     - no token is required, so `tokenEnv` is irrelevant here.
+    #
+    #   MUTUALLY EXCLUSIVE with `baseUrl` — setting both is refused at config load (CB-542). On the
+    #   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`
+    #   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.
     # 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)

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

@@ -9,6 +9,10 @@ import dev.ltms.bridged.peer.Capability;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+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;
@@ -230,7 +234,7 @@ public final class ClaudeCodeLauncher extends HerdrPeerLauncher {
         // 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()));
+        List<String> argv = mutableArgv(argvWithBridge(cfg, spec));
         String agentSessionId = applySessionIdentity(argv, spec.sessionName(), spec.resumeSessionId());
         return new Launch(workerEnv, argvWithModel(argv, cfg), agentSessionId);
     }
@@ -265,13 +269,31 @@ public final class ClaudeCodeLauncher extends HerdrPeerLauncher {
     }
 
     /**
-     * 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.
+     * The launch argv, plus an inline {@code --mcp-config} when {@code worker.mcpUrl} is set, the
+     * CB-617 charter flags, and {@code --agent <role>} when the role has an agent-definition file
+     * under the worker's cwd. Neither touches the profile's config; all are pure command-line flags.
+     * This inline-flag mount is Claude Code specific — other adapters mount MCP and instructions
+     * their own way.
+     *
+     * <p>CB-617: the role charter is operator-authored and often multi-line, so it can never be a
+     * single inline argv element — herdr refuses to shell-encode a multi-line argument
+     * ({@code invalid_agent_argument}). It is written to a temp file instead and mounted with
+     * {@code --append-system-prompt-file}, which this host confirms Claude Code accepts for a
+     * multi-line file.
+     *
+     * <p>CB-618: Claude Code refuses to start when BOTH {@code --append-system-prompt} and
+     * {@code --append-system-prompt-file} are on the command line ("Cannot use both ... Please use
+     * only one"), so the two charters can never travel on separate flags. When both are present they
+     * are concatenated into the one file, role charter first and reply charter last — last is where
+     * the reply rule must sit, because it is the rule that must survive. When only the reply charter
+     * is present it keeps its proven inline {@code --append-system-prompt} delivery, which is also
+     * the only form that reaches a member with no repo checkout.
      */
-    private List<String> argvWithBridge(BridgedConfig.Profile cfg, String charter) {
-        if (!cfg.hasMcp() && charter == null) {
+    private List<String> argvWithBridge(BridgedConfig.Profile cfg, LaunchSpec spec) {
+        String roleCharter = nonBlank(spec.roleCharter());
+        String replyCharter = nonBlank(spec.replyCharter());
+        Path agentFile = agentDefinitionFile(spec.cwd(), spec.role(), ".claude", "agents");
+        if (!cfg.hasMcp() && roleCharter == null && replyCharter == null && agentFile == null) {
             return cfg.argv();
         }
         List<String> argv = mutableArgv(cfg.argv());
@@ -281,13 +303,45 @@ public final class ClaudeCodeLauncher extends HerdrPeerLauncher {
             argv.add("--mcp-config");
             argv.add(mcpJson);
         }
-        if (charter != null) {
+        if (roleCharter != null) {
+            String combined = replyCharter == null ? roleCharter : roleCharter + "\n\n" + replyCharter;
+            argv.add("--append-system-prompt-file");
+            argv.add(writeCharterFile(combined).toString());
+        } else if (replyCharter != null) {
             argv.add("--append-system-prompt");
-            argv.add(charter);
+            argv.add(replyCharter);
+        }
+        if (agentFile != null) {
+            argv.add("--agent");
+            argv.add(spec.role().wireName());
         }
         return argv;
     }
 
+    /** {@code s}, or {@code null} when {@code s} is null/blank — the charter-presence test used above. */
+    private static String nonBlank(String s) {
+        return (s == null || s.isBlank()) ? null : s;
+    }
+
+    /**
+     * Write the role charter to a fresh temp file so it can be mounted with
+     * {@code --append-system-prompt-file} instead of riding inline in argv (CB-617). Best-effort
+     * cleaned via {@code deleteOnExit} — the same disposable-worker-config cleanup
+     * {@link OpenCodeLauncher#writeConfig} already uses for its charter file, since the process that
+     * reads this file (the spawned peer) outlives this JVM call and there is no spawn-scoped teardown
+     * hook to delete it synchronously.
+     */
+    private static Path writeCharterFile(String charterText) {
+        try {
+            Path file = Files.createTempFile("bridged-role-charter-", ".md");
+            Files.writeString(file, charterText);
+            file.toFile().deleteOnExit();
+            return file;
+        } catch (IOException e) {
+            throw new UncheckedIOException("cannot write role charter temp file", e);
+        }
+    }
+
     /**
      * Pin the model on the command line as well as in {@code ANTHROPIC_MODEL} (CB-533).
      *

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

@@ -17,6 +17,8 @@ import dev.ltms.bridged.peer.SpawnRequest;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
+import java.nio.file.Files;
+import java.nio.file.Path;
 import java.security.SecureRandom;
 import java.util.ArrayList;
 import java.util.Collection;
@@ -279,8 +281,35 @@ public abstract class HerdrPeerLauncher implements PeerLauncher {
         }
     }
 
-    /** All per-spawn values adapters may need, including the base-composed effective charter. */
-    protected record LaunchSpec(String sessionName, String resumeSessionId, MemberRole role, String charter) {
+    /**
+     * All per-spawn values adapters may need.
+     *
+     * <p>{@code charter} is the base-composed effective charter (role charter, then the reply
+     * charter, joined by a blank line) — kept for an adapter that mounts both as one blob (opencode
+     * writes it to a single file) and as the exact input {@link CharterReceipt#compose} fingerprints.
+     * {@code roleCharter} and {@code replyCharter} are the same text split back into its two parts
+     * (CB-617), for an adapter that must deliver them differently: the role charter is
+     * operator-authored and often multi-line, so it cannot travel as an inline argv element (herdr
+     * refuses to shell-encode a multi-line argument); the reply charter is always one line and is
+     * demonstrated to encode, so it may still go inline. {@code cwd} is the spawn's resolved working
+     * directory (CB-112), needed to look up a role's agent-definition file before the peer starts.
+     */
+    protected record LaunchSpec(String sessionName, String resumeSessionId, MemberRole role, String charter,
+                                String roleCharter, String replyCharter, String cwd) {
+    }
+
+    /**
+     * The role's agent-definition file under {@code <cwd>/<dir1>/<dir2>/<role>.md}, or {@code null}
+     * when absent or inapplicable (no role, no cwd, or the file does not exist) — CB-617. A member
+     * whose role has no such file must still spawn, so this is a lookup, never a requirement: the
+     * caller passes {@code --agent <role>} only when the return value is non-null.
+     */
+    protected static Path agentDefinitionFile(String cwd, MemberRole role, String dir1, String dir2) {
+        if (role == null || cwd == null || cwd.isBlank()) {
+            return null;
+        }
+        Path candidate = Path.of(cwd, dir1, dir2, role.wireName() + ".md");
+        return Files.isRegularFile(candidate) ? candidate : null;
     }
 
     // --- profile surface -----------------------------------------------------------------------
@@ -387,9 +416,10 @@ public abstract class HerdrPeerLauncher implements PeerLauncher {
         // 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);
+        String cwd = resolveCwd(requestedCwd, cfg, callerCwd);
         try {
-            Launch launch = buildLaunch(cfg, new LaunchSpec(sessionName, resumeSessionId, role, charter));
-            String cwd = resolveCwd(requestedCwd, cfg, callerCwd);
+            Launch launch = buildLaunch(cfg, new LaunchSpec(sessionName, resumeSessionId, role, charter,
+                    roleCharter, replyCharter, cwd));
             Agent agent = cfg.tabPlacement()
                     ? spawnInTab(cfg, launch.env(), launch.argv(), cwd, role, liveFleet)
                     : spawnAsPane(cfg, launch.env(), launch.argv(), cwd, charter);

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

@@ -208,8 +208,24 @@ public final class OpenCodeLauncher extends HerdrPeerLauncher {
             workerEnv.put("OPENCODE_CONFIG", writeConfig(cfg, spec.charter()).toString());
         }
         applyGitToken(workerEnv, cfg);
-        return new Launch(workerEnv,
-                argvWithResume(argvWithModel(argvWithAuto(cfg), cfg), spec.resumeSessionId()));
+        List<String> argv = argvWithResume(argvWithModel(argvWithAuto(cfg), cfg), spec.resumeSessionId());
+        return new Launch(workerEnv, argvWithAgent(argv, spec));
+    }
+
+    /**
+     * The launch argv plus, when the role has an agent-definition file under the worker's cwd,
+     * opencode's {@code --agent <role>} flag (CB-617). A role with no such file gets nothing added —
+     * the member must still spawn.
+     */
+    private List<String> argvWithAgent(List<String> argv, LaunchSpec spec) {
+        Path agentFile = agentDefinitionFile(spec.cwd(), spec.role(), ".opencode", "agent");
+        if (agentFile == null) {
+            return argv;
+        }
+        List<String> withAgent = mutableArgv(argv);
+        withAgent.add("--agent");
+        withAgent.add(spec.role().wireName());
+        return withAgent;
     }
 
     /**

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

@@ -15,8 +15,11 @@ 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 org.slf4j.LoggerFactory;
 
+import java.nio.file.Files;
+import java.nio.file.Path;
 import java.util.List;
 import java.util.Map;
 import java.util.Set;
@@ -106,10 +109,24 @@ class ClaudeCodeLauncherTest {
                 "the operator's own args are preserved, in order, ahead of the model flag");
     }
 
+    // --- CB-617/CB-618: both charters travel in ONE file; the reply charter alone stays inline ---
+
+    /**
+     * CB-617: herdr refuses to shell-encode a multi-line inline argv argument
+     * ({@code invalid_agent_argument}) — the exact failure this reproduced on profile {@code opus}.
+     * The role charter is operator-authored and often multi-line, so it must never appear as an argv
+     * element.
+     *
+     * <p>CB-618: and Claude Code itself refuses to start when both {@code --append-system-prompt} and
+     * {@code --append-system-prompt-file} are given ("Cannot use both ... Please use only one"), so
+     * the reply charter cannot ride inline alongside a role charter either. Both go in the one file,
+     * reply charter last. This drives the real launcher entry point ({@code spawn}), the same path a
+     * live spawn takes — not the argv builder in isolation.
+     */
     @Test
-    void appendsTheBaseComposedRoleAndReplyCharter() {
+    void bothChartersTravelInOneFileAndNeverOnBothFlags() {
         FakeHerdr herdr = new FakeHerdr();
-        String roleCharter = "You review changes.";
+        String roleCharter = "You review changes.\nLine two.\nLine three.";
         BridgedConfig.Profile cfg = new BridgedConfig.Profile(
                 "sonnet", "http://gx00.gw:8000", null, null, "BRIDGED_WORKER_TOKEN",
                 List.of("claude"), "tab", "bridged-workers", "w #{n}",
@@ -121,11 +138,20 @@ class ClaudeCodeLauncherTest {
         svc.spawn(new SpawnRequest("sonnet", null, null, null, null, MemberRole.REVIEWER));
 
         List<String> args = spawnedArgs(herdr);
-        int flag = args.indexOf("--append-system-prompt");
-        assertEquals(1, args.stream().filter("--append-system-prompt"::equals).count(),
-                "the composed charter is passed once");
-        assertEquals(roleCharter + "\n\n" + HerdrPeerLauncher.REPLY_CHARTER, args.get(flag + 1),
-                "the role charter comes first and the reply rule comes last");
+        assertTrue(args.stream().noneMatch(a -> a.contains("\n")),
+                "no argv element may be multi-line — herdr cannot shell-encode one: " + args);
+
+        int fileFlag = args.indexOf("--append-system-prompt-file");
+        assertTrue(fileFlag >= 0, "the role charter is mounted via --append-system-prompt-file: " + args);
+        assertFalse(args.contains("--append-system-prompt"),
+                "CB-618: Claude Code refuses to start with both flags — the reply charter must not "
+                        + "ride inline beside a role charter: " + args);
+        assertDoesNotThrow(() -> {
+            String written = Files.readString(Path.of(args.get(fileFlag + 1)));
+            assertTrue(written.startsWith(roleCharter), "the file opens with the role charter: " + written);
+            assertTrue(written.endsWith(HerdrPeerLauncher.REPLY_CHARTER),
+                    "the reply charter is last — it is the rule that must survive: " + written);
+        }, "the --append-system-prompt-file path must be a readable file");
     }
 
     @Test
@@ -135,11 +161,13 @@ class ClaudeCodeLauncherTest {
 
         svc.spawn(new SpawnRequest("sonnet", null, null, null, null, MemberRole.DEV));
 
-        assertFalse(spawnedArgs(herdr).contains("--append-system-prompt"));
+        List<String> args = spawnedArgs(herdr);
+        assertFalse(args.contains("--append-system-prompt"));
+        assertFalse(args.contains("--append-system-prompt-file"));
     }
 
     @Test
-    void profileWithoutMcpStillGetsItsRoleCharter() {
+    void profileWithoutMcpStillGetsItsRoleCharterAsAFileNotInline() {
         FakeHerdr herdr = new FakeHerdr();
         String roleCharter = "You design changes.";
         ClaudeCodeLauncher svc = labelService(herdr, () -> fleet(Map.of("architect", roleCharter), null));
@@ -147,12 +175,52 @@ class ClaudeCodeLauncherTest {
         svc.spawn(new SpawnRequest("sonnet", null, null, null, null, MemberRole.ARCHITECT));
 
         List<String> args = spawnedArgs(herdr);
-        int flag = args.indexOf("--append-system-prompt");
+        int flag = args.indexOf("--append-system-prompt-file");
         assertTrue(flag >= 0, "a role charter does not need an MCP mount");
-        assertEquals(roleCharter, args.get(flag + 1));
+        assertDoesNotThrow(() -> assertEquals(roleCharter,
+                        Files.readString(Path.of(args.get(flag + 1)))),
+                "the file holds the role charter");
+        assertFalse(args.contains("--append-system-prompt"), "no reply charter without an MCP mount");
         assertFalse(args.contains("--mcp-config"));
     }
 
+    // --- CB-617: --agent <role> when the role has an agent-definition file --------------------
+
+    @Test
+    void agentFlagIsPassedWhenTheRoleAgentDefinitionFileExists(@TempDir Path cwd) throws Exception {
+        FakeHerdr herdr = new FakeHerdr();
+        Path agentsDir = Files.createDirectories(cwd.resolve(".claude/agents"));
+        Files.writeString(agentsDir.resolve("reviewer.md"), "---\nname: reviewer\n---\nBe a reviewer.");
+        BridgedConfig.Profile cfg = new BridgedConfig.Profile(
+                "sonnet", "http://gx00.gw:8000", null, null, "BRIDGED_WORKER_TOKEN",
+                List.of("claude"), "tab", "bridged-workers", "w #{n}", null, null, null);
+        ClaudeCodeLauncher svc = new ClaudeCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
+                new SubscriptionGuard(Set.of("gx00.gw")), Map.of(cfg.profile(), cfg), cfg.profile(), _ -> null);
+
+        svc.spawn(new SpawnRequest("sonnet", cwd.toString(), null, null, null, MemberRole.REVIEWER));
+
+        List<String> args = spawnedArgs(herdr);
+        int flag = args.indexOf("--agent");
+        assertTrue(flag >= 0, "--agent is passed when the role's agent file exists: " + args);
+        assertEquals("reviewer", args.get(flag + 1));
+    }
+
+    @Test
+    void noAgentFlagWhenTheRoleAgentDefinitionFileIsAbsent(@TempDir Path cwd) {
+        FakeHerdr herdr = new FakeHerdr();
+        BridgedConfig.Profile cfg = new BridgedConfig.Profile(
+                "sonnet", "http://gx00.gw:8000", null, null, "BRIDGED_WORKER_TOKEN",
+                List.of("claude"), "tab", "bridged-workers", "w #{n}", null, null, null);
+        ClaudeCodeLauncher svc = new ClaudeCodeLauncher(new AgentControl(herdr), new WorkspaceControl(herdr),
+                new SubscriptionGuard(Set.of("gx00.gw")), Map.of(cfg.profile(), cfg), cfg.profile(), _ -> null);
+
+        PeerHandle handle = svc.spawn(new SpawnRequest("sonnet", cwd.toString(), null, null, null, MemberRole.REVIEWER));
+
+        assertNotNull(handle, "the member still spawns with no agent-definition file");
+        assertFalse(spawnedArgs(herdr).contains("--agent"),
+                "no --agent flag when the role has no agent-definition file");
+    }
+
     private ClaudeCodeLauncher multiProfile(FakeHerdr herdr) {
         BridgedConfig.Profile gx10 = new BridgedConfig.Profile("gx10", "http://gx10.gw:8000", "coder",
                 null, "BRIDGED_WORKER_TOKEN", List.of("claude"), "tab", "bridged-workers", "w #{n}", null, null, null);

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

@@ -8,6 +8,7 @@ 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.MemberRole;
 import dev.ltms.bridged.peer.PeerHandle;
 import dev.ltms.bridged.peer.PeerUnreachableException;
 import dev.ltms.bridged.peer.SpawnRequest;
@@ -208,6 +209,35 @@ class OpenCodeLauncherTest {
                 "--auto is unconditional: a model-less worker still must never block on approval");
     }
 
+    // --- CB-617: --agent <role> when the role has an agent-definition file --------------------
+
+    @Test
+    void agentFlagIsPassedWhenTheRoleAgentDefinitionFileExists(@TempDir Path root) throws Exception {
+        FakeHerdr herdr = new FakeHerdr();
+        Path agentsDir = Files.createDirectories(root.resolve(".opencode/agent"));
+        Files.writeString(agentsDir.resolve("dev.md"), "You are a dev.");
+        OpenCodeLauncher svc = service(herdr, root, opencodeCfg(null, null, null));
+
+        svc.spawn(new SpawnRequest(null, root.toString(), null, null, null, MemberRole.DEV));
+
+        List<String> args = startArgs(herdr);
+        int flag = args.indexOf("--agent");
+        assertTrue(flag >= 0, "--agent is passed when the role's agent file exists: " + args);
+        assertEquals("dev", args.get(flag + 1));
+    }
+
+    @Test
+    void noAgentFlagWhenTheRoleAgentDefinitionFileIsAbsent(@TempDir Path root) {
+        FakeHerdr herdr = new FakeHerdr();
+        OpenCodeLauncher svc = service(herdr, root, opencodeCfg(null, null, null));
+
+        PeerHandle handle = svc.spawn(new SpawnRequest(null, root.toString(), null, null, null, MemberRole.DEV));
+
+        assertNotNull(handle, "the member still spawns with no agent-definition file");
+        assertFalse(startArgs(herdr).contains("--agent"),
+                "no --agent flag when the role has no agent-definition file");
+    }
+
     @Test
     void injectsForgeTokenWhenProfileGrantsIt(@TempDir Path root) {
         FakeHerdr herdr = new FakeHerdr();

docs/MCP-Contract.md

@@ -1,9 +1,24 @@
 # MCP Contract — `bridged`'s unified gateway
 
-> **Status:** 🟡 Design (2026-07-14). Greenfield — no MCP code exists yet; the pom carries
-> only Javalin/Jackson. This page defines the tool surface that CB-104 and its followers
-> implement. It supersedes nothing; it fills the "MCP server face" left open by the
-> [Architecture](1-Architecture) page.
+> **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:
+>
+> - **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}`.
+>
+> **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.
+>
+> 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)

docs/Team.md

@@ -17,7 +17,7 @@ who the workers are, how the lead picks one, and how it runs many at once.
 - **Workers** — a herd of `claude` panes in herdr, each an addressable `bridged` 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://ollama.ltms.dev`) — bulk, cheap, or
+  - **Local workers** (`ANTHROPIC_BASE_URL=https://llm.ltms.dev/anthropic`) — bulk, cheap, or
     embarrassingly parallel subtasks.
 
 Every worker is still a *real Claude Code process* (inherits `CLAUDE.md`, hooks, skills,
@@ -35,7 +35,7 @@ flowchart TB
     WL1["w-local-1<br/>ANTHROPIC_BASE_URL set"]
     WL2["w-local-2<br/>ANTHROPIC_BASE_URL set"]
     ANT["api.anthropic.com<br/>(Pro/Max)"]
-    OLL["ollama.ltms.dev<br/>(local model)"]
+    OLL["llm.ltms.dev<br/>(gateway to the local model)"]
 
     LEAD -->|"blocking POST /message (target role)"| BD
     BD -->|"Unix socket · send_text · events.subscribe"| HERDR

opencode.json

@@ -4,7 +4,7 @@
     "CLAUDE.md"
   ],
   "mcp": {
-    "bridged": {
+    "fleetd": {
       "type": "remote",
       "url": "http://127.0.0.1:8765/mcp",
       "enabled": true