diff --git a/fleetd/src/test/java/dev/ltms/fleet/config/FleetConfigWithDefaultsPreservesEveryComponentTest.java b/fleetd/src/test/java/dev/ltms/fleet/config/FleetConfigWithDefaultsPreservesEveryComponentTest.java
new file mode 100644
index 0000000..1f710a0
--- /dev/null
+++ b/fleetd/src/test/java/dev/ltms/fleet/config/FleetConfigWithDefaultsPreservesEveryComponentTest.java
@@ -0,0 +1,192 @@
+package dev.ltms.fleet.config;
+
+import org.junit.jupiter.api.Test;
+
+import java.lang.reflect.Constructor;
+import java.lang.reflect.RecordComponent;
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.LinkedHashMap;
+import java.util.List;
+import java.util.Locale;
+import java.util.Map;
+import java.util.Objects;
+import java.util.Set;
+import java.util.TreeSet;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+/**
+ * Guards against a "defect factory" built into this file's own established pattern, found live
+ * while building the (parked) idle-sleep-guard PR: every time a component is added to
+ * {@link FleetConfig}, the record grows by one arg AND a new back-compat constructor is added at
+ * the OLD arity, so existing callers keep compiling. That is correct and required — see the
+ * constructor ladder just below the record header. But {@link #withDefaults()}'s own {@code return
+ * new FleetConfig(...)} call sits in this same file, written at a literal argument count. The very
+ * next time a component is added, the freshly-added back-compat constructor at the OLD arity
+ * silently captures that stale call, because it is now a legal overload at that arg count too. It
+ * compiles. Every other test passes, because nothing else exercises the new field. The new
+ * component is defaulted away — {@code null}, or whatever that back-compat overload defaults it to
+ * — on every {@link FleetConfig#load}. Measured, not theoretical: this exact sequence happened
+ * live when the {@code idleSleepGuard} component was added on a sibling branch; it was caught only
+ * because that branch's own new tests happened to assert on the new field's value.
+ *
+ * <p>This test proves the opposite property, and does it in a way that survives the next field
+ * being added without being rewritten: reflectively enumerate {@link FleetConfig}'s own record
+ * components (never a hardcoded count — the arity is exactly what changes over time), build one
+ * config through the true canonical constructor with a real, distinctive, non-null value in EVERY
+ * component (reusing the exact reflective-construction pattern
+ * {@link ConfigRefTopLevelReportingCoverageTest} already established for this file:
+ * {@code getDeclaredConstructor(exact record-component types)}, which resolves the canonical
+ * constructor by its true shape, not by binding to whichever overload happens to match arg count —
+ * the same way Jackson resolves it), call the real {@link FleetConfig#withDefaults()}, and assert
+ * every one of those values survives unchanged.
+ *
+ * <p>Why this is a valid check for every component, not just some: {@link #withDefaults()}'s own
+ * comments document that it only ever REPLACES a component when the incoming value is {@code null}
+ * (or blank, for {@code placement}) — {@code broker}/{@code primary}/{@code leadHeartbeat}/
+ * {@code configReload}/{@code coordinator}/{@code worktreeGroup}/{@code memberLoginShell} are left
+ * as-is unconditionally, and {@code bind}/{@code guard}/{@code lifecycle}/{@code auth}/
+ * {@code fleet}/{@code quarantineCooldownSeconds}/{@code memberCredentials}/{@code placement} are
+ * replaced only on null/blank input. A value that is never null or blank going in must therefore
+ * never change coming out, for every current component. No exclusion is needed today.
+ *
+ * <p>{@link #EXCLUDED_FROM_SURVIVAL_CHECK} exists anyway, kept deliberately empty and size-pinned
+ * by {@link #exclusionListSizeIsPinned()}: a future component that {@code withDefaults()} is
+ * <em>documented</em> to transform unconditionally (unlike every field today) would legitimately
+ * need one. Pinning the size at 0 means growing that set to make a failure go away is itself a
+ * visible diff to this test, not a silent one — a checker that can be silenced by adding to its
+ * own escape hatch is not a checker.
+ */
+class FleetConfigWithDefaultsPreservesEveryComponentTest {
+
+    private static final RecordComponent[] COMPONENTS = FleetConfig.class.getRecordComponents();
+
+    /** See the class javadoc — deliberately empty today; grow it only with a matching justification. */
+    private static final Set<String> EXCLUDED_FROM_SURVIVAL_CHECK = Set.of();
+
+    /** One real, distinctive, non-null (non-blank where blankness would mean "unset") value per component. */
+    private static Map<String, Object> baseValues() {
+        Map<String, Object> v = new LinkedHashMap<>();
+        v.put("bind", new FleetConfig.Bind("127.0.0.1", 8765));
+        v.put("herdrSocket", "~/.config/herdr/guard.sock");
+        v.put("memberHerdrSocket", "~/.config/herdr/member-guard.sock");
+        v.put("profiles", Map.of("sonnet", minimalProfile("sonnet")));
+        v.put("guard", new FleetConfig.Guard(List.of("host-guard")));
+        v.put("worktreeRoot", "/wt/guard");
+        v.put("lifecycle", new FleetConfig.Lifecycle(300, 5, 30, true));
+        v.put("spawnReadyTimeoutMs", 12_345);
+        v.put("spawnReadyPollMs", 234);
+        v.put("broker", new FleetConfig.Broker("amqp://guard", null, 7));
+        v.put("primary", new FleetConfig.Primary("term-guard", 4, 4000));
+        v.put("fleet", new FleetConfig.Fleet(
+                Map.of("opus", new FleetConfig.Leader("sonnet", "lead: opus-guard", 1, null, 10,
+                        "claude", null, null, null)),
+                Map.of(), Map.of(), Map.of(), Map.of(), "{role}: {profile} #{n}"));
+        v.put("leadHeartbeat", new FleetConfig.LeadHeartbeat(301, 61_000L, 4));
+        v.put("health", new FleetConfig.Health(true, 31, 601, 61, null));
+        v.put("placement", "round-robin");
+        v.put("auth", new FleetConfig.Auth("loopback-trust", null));
+        v.put("configReload", new FleetConfig.ConfigReload(true, 11));
+        v.put("quarantineCooldownSeconds", 1801);
+        v.put("memberCredentials", new FleetConfig.MemberCredentials(
+                FleetConfig.MemberCredentials.POLICY_DENY_BY_DEFAULT,
+                List.of("git"), List.of("git", "ssh"), null));
+        v.put("coordinator", new FleetConfig.Coordinator("amqp://coord-guard", null, "self-guard", 3));
+        v.put("worktreeGroup", "group-guard");
+        v.put("memberLoginShell", "/bin/zsh");
+        assertNamesMatchComponents(v);
+        return v;
+    }
+
+    /** A minimal, otherwise-null {@link FleetConfig.Profile} — just enough to name one in a map. */
+    private static FleetConfig.Profile minimalProfile(String name) {
+        return new FleetConfig.Profile(name, null, null, null, null, null, null, null, null, null,
+                null, null, null, null, null, null, null, null, null, null, null, null, null, null,
+                null, null);
+    }
+
+    /**
+     * Guards {@link #baseValues()} itself against drifting from the record's real shape — the same
+     * assurance {@link ConfigRefTopLevelReportingCoverageTest} already relies on. This is what makes
+     * "no hardcoded arity" true in practice: forgetting to add a new component here fails this
+     * assertion by name, rather than silently checking one component fewer than the record has.
+     */
+    private static void assertNamesMatchComponents(Map<String, Object> values) {
+        Set<String> names = new TreeSet<>();
+        for (RecordComponent rc : COMPONENTS) {
+            names.add(rc.getName());
+        }
+        assertEquals(names, new TreeSet<>(values.keySet()),
+                "this test's value map has drifted from FleetConfig's actual top-level components — "
+                        + "update baseValues() alongside the record");
+    }
+
+    /**
+     * Builds a {@link FleetConfig} through the TRUE canonical constructor — resolved by the record's
+     * own component types, not by argument count — so this never accidentally exercises a
+     * back-compat overload the way a literal {@code new FleetConfig(...)} call risks doing.
+     */
+    private static FleetConfig configOf(Map<String, Object> values) throws ReflectiveOperationException {
+        Class<?>[] types = Arrays.stream(COMPONENTS).map(RecordComponent::getType).toArray(Class<?>[]::new);
+        Object[] args = Arrays.stream(COMPONENTS).map(rc -> values.get(rc.getName())).toArray();
+        Constructor<FleetConfig> ctor = FleetConfig.class.getDeclaredConstructor(types);
+        return ctor.newInstance(args);
+    }
+
+    @Test
+    void exclusionListSizeIsPinned() {
+        assertEquals(0, EXCLUDED_FROM_SURVIVAL_CHECK.size(),
+                "EXCLUDED_FROM_SURVIVAL_CHECK grew from 0 — every entry needs a justification in "
+                        + "this test class's javadoc AND this assertion re-pinned to the new size; a "
+                        + "growing exclusion list that silences failures on its own is not a guard");
+    }
+
+    /**
+     * The mutation this is built to catch: make {@code withDefaults()}'s final constructor call
+     * literal at some arg count, add one more component to the record with a new back-compat
+     * constructor at the old arity, and the stale call silently rebinds. Every component here is
+     * real and non-null (non-blank for the one String — {@code placement} — where blank has
+     * meaning), so none of it should be replaced by {@code withDefaults()}; any component that
+     * comes back different was silently dropped.
+     */
+    @Test
+    void everyComponentGivenARealValueSurvivesWithDefaults() throws ReflectiveOperationException {
+        Map<String, Object> base = baseValues();
+        FleetConfig config = configOf(base);
+        FleetConfig defaulted = config.withDefaults();
+
+        List<String> dropped = new ArrayList<>();
+        int checked = 0;
+        for (RecordComponent rc : COMPONENTS) {
+            String name = rc.getName();
+            if (EXCLUDED_FROM_SURVIVAL_CHECK.contains(name)) {
+                continue;
+            }
+            checked++;
+            Object expected = base.get(name);
+            Object actual;
+            try {
+                actual = rc.getAccessor().invoke(defaulted);
+            } catch (ReflectiveOperationException e) {
+                throw new RuntimeException("failed to read FleetConfig." + name + "()", e);
+            }
+            if (!Objects.equals(expected, actual)) {
+                dropped.add(String.format(Locale.ROOT,
+                        "%s: withDefaults() was given a real, non-null value (%s) for '%s' but "
+                                + "returned %s — a component silently dropped by withDefaults(), the "
+                                + "shape of the defect this test exists to catch (its final "
+                                + "\"return new FleetConfig(...)\" call binding to a back-compat "
+                                + "constructor instead of the true canonical one)",
+                        name, expected, name, actual));
+            }
+        }
+
+        System.out.printf(Locale.ROOT,
+                "FleetConfig.withDefaults() component-survival coverage — %d components, %d checked, "
+                        + "%d excluded, %d survived%n",
+                COMPONENTS.length, checked, EXCLUDED_FROM_SURVIVAL_CHECK.size(), checked - dropped.size());
+        assertEquals(List.of(), dropped,
+                "withDefaults() silently dropped these real, given components: " + dropped);
+    }
+}