diff --git a/BUILD-LOG.md b/BUILD-LOG.md
index b281ea8c9..64bffa95a 100644
--- a/BUILD-LOG.md
+++ b/BUILD-LOG.md
@@ -4256,3 +4256,11 @@ Before (a0080da8): no test covered a directory pin, a symlink default or a pin w
 After (candidate, uncommitted packet agents/rocko/work/queue-72/, candidate manifest e6531a53): one test in host.test.mjs, one in claude-session.test.mjs, and mutants.sh. All six mutants are killed on the candidate and all six survive on the base. Deviation from the brief: a directory named claude on PATH can't reach the default pin, because bash's command -v skips directories. The test asserts the real behaviour: no pin, with exit 0. No change to bus-service.sh or adapter.sh.
 
 Gate (Rocko, scratch worktree a0080da8 plus the candidate, sequential): cli 113/113, harness 58/58 with 0 skipped; test-auth 15, test-conductor 17, test-config 24, test-discord 66, test-extension-package 18, test-foundation 44, test-queue 27, test-release 14, test-task 98, all with 0 failed (Docker). Removed the rocko-r72_default network. No commits, no pushes, no tokens read.
+
+### 2026-10-11 — Rocko, row 73 round 1 candidate: drop the coder's task.reassign grant (#1567)
+
+Before (324c064b): `roles/coder.json` grants `crossRole: ["task.reassign", "task.scope.change"]`. The broker classes a coder's `task.reassign` as cross-role and routes its decision to pm, and a coder with a tracker gets a `task_reassign` tool. Yet the task verb refuses every non-pm `task.reassign` with `field-writer` before authority is checked (lead decision 90), so the grant can never succeed. The mutant `broker-unlisted-cross` (an unlisted action becomes cross-role) survives the base tests.
+
+After (candidate, uncommitted packet agents/rocko/work/queue-73/, candidate manifest bf4e5ed6, packet manifest 41ffcd5c, 53 files): the coder's `crossRole` is `["task.scope.change"]`. Its `task.reassign` is gated and routes to the human, and it gets no `task_reassign` tool. New or changed tests are in business role and resolve, bus broker (a test built on the shipped roles), harness tools and tasks verbs, and the `world.mjs` fixture now matches the shipped role. The narrowing test moved from the coder to the pm, because with one cross-role action left the old test passes under `resolve-cross-unnarrowed`. All 5 mutants are killed on the candidate. No change to broker.mjs or PM_VERBS. Not changed, named in BUILD.md for Sage: `packages/business/README.md` line 37, `packages/tasks/README.md` lines 193-196 and `packages/tasks/live/run.mjs` line 32 still show the old grant.
+
+Gate (Rocko, scratch worktree 324c064b plus the candidate, sequential, 23:09:50Z-23:14:34Z): tasks 71, bus 81, business 60, harness 58 (0 skipped), cli 112, webui 50, seat 27 node tests; test-auth 15, test-conductor 17, test-config 24, test-discord 66, test-extension-package 18, test-foundation 44, test-queue 27, test-release 14, test-task 98, all 0 failed (Docker). Removed the rocko-r73_default network. Queue revs 533-534; the request failed without a credential, as expected. No commits, no pushes, no tokens read.
diff --git a/agents/rocko/work/queue-73/BUILD.md b/agents/rocko/work/queue-73/BUILD.md
new file mode 100644
index 000000000..6330dc86d
--- /dev/null
+++ b/agents/rocko/work/queue-73/BUILD.md
@@ -0,0 +1,140 @@
+# Row 73 (#1567): drop the coder's task.reassign grant, candidate packet, round 1
+
+Author: Rocko. Reviewer: Dewey. Brief:
+`docs/plans/2026-10-11_claude-pin-and-coder-grant.md`, section "Coder role:
+drop the task.reassign grant the broker always refuses". Base: `324c064b`
+(`base.txt`). At 23:16Z the checkout HEAD `f6b9ea4e` has no change since the
+base to any candidate path or to `packages/{bus,business,tasks}/src` or
+`packages/harness/src/tools.mjs`. `build.patch` applies to it cleanly. The
+packet is uncommitted. I made no commits, pushes or Gitea calls and read no
+token or private binding.
+
+## Files (`files.txt`, `candidate-manifest.sha256`)
+
+| File | Change |
+|---|---|
+| `roles/coder.json` | `crossRole`: `["task.reassign", "task.scope.change"]` → `["task.scope.change"]`. Nothing else in the file changes. |
+| `packages/business/tests/role.test.mjs` | "shipped authority follows the note's table": asserts that the coder's `crossRole` is exactly `["task.scope.change"]` |
+| `packages/business/tests/resolve.test.mjs` | "classify": the coder's `task.reassign` is now `gated` and its `task.scope.change` is `cross`. "limits.authority narrows cross-role actions too" moves from the coder to the pm (see below). |
+| `packages/bus/tests/broker.test.mjs` | +1 test built on the shipped role files: "the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role" |
+| `packages/harness/tests/tools.test.mjs` | +1 test: "a coder with a tracker gets its own task tools and no task_reassign" |
+| `packages/tasks/tests/world.mjs` | the fixture's coder `crossRole` now matches the shipped role |
+| `packages/tasks/tests/verbs.test.mjs` | in "assign and reassign …": a coder's `task.reassign` refuses `field-writer` and records no `action.allowed` |
+| `agents/rocko/work/queue-73/mutants.sh` | the mutant rerun: `mutants.sh <tree>` |
+
+No change to `broker.mjs`, `PM_VERBS`, the slice 1 field table or any other
+role's grants.
+
+### Why the narrowing test moved to the pm
+
+Before this change, the test limited the coder to `task.scope.change` and
+watched `task.reassign` drop out of `crossRole`. With one cross-role action
+left, that limit drops nothing from `crossRole`. I checked this: with the new
+role file and the old test, the mutant `resolve-cross-unnarrowed` (crossRole
+ignores `limits.authority`) passes the test (`out/old-narrow-test-under-mutant.txt`).
+The pm still holds two cross-role actions. The test now limits the pm to
+`task.create`, `task.close` and `task.priority.change`. It checks that
+`crossRole` keeps `task.priority.change` and drops `task.scope.change`, and
+that `withinRole` narrows too.
+
+## Every place that read the old grant, and what it says now
+
+Probe (`out/probe-base.txt`, `out/probe-candidate.txt`): the shipped roles
+through `validateBusinessDocument`, `resolveInstance`, `busBusiness` and a
+real `Broker`, as `bootConfig` assembles them.
+
+| Reader | Before (base) | After (candidate) |
+|---|---|---|
+| `business/src/resolve.mjs` `resolveInstance` | coder `crossRole` `[task.reassign, task.scope.change]` | `[task.scope.change]`. The resolved digest of every coder instance changes too. Only the bundle manifest records it, and nothing compares it. |
+| `business/src/resolve.mjs` `classify` | `cross` | `gated` |
+| `bus/src/broker.mjs` `#classification` (via `cli/src/config.mjs` `bootConfig` → `busBusiness`) | `cross-role`; `decision.raise` routes to the domain arbiter, `pm` by default (`cto` for technical) | `gated`; routes to `human` |
+| `harness/src/tools.mjs` `typedTools`, with a tracker | the coder gets a `task_reassign` tool | no `task_reassign` tool; `task_scope_change` and `task_update_assigned` remain |
+| `harness/src/tools.mjs` `untypedActions` → `bundle.mjs` prompt, without a tracker | the prompt lists `task.reassign` among "Actions you hold that have no bus tool yet" | not listed |
+| `tasks/src/verbs.mjs` `PM_VERBS` | `field-writer` before authority | unchanged: `field-writer` before authority |
+| `tasks/tests/world.mjs` fixture | copied the old grant | matches the shipped role |
+| business, bus, harness and tasks tests | see Files | see Files |
+
+Readers I did not change (outside the brief's files; Sage's call):
+
+- `packages/business/README.md`, lines 30-43: the version 2 example is the
+  shipped coder file and still shows `"crossRole": ["task.reassign",
+  "task.scope.change"]`. It goes stale with this change. Suggested fix: drop
+  `"task.reassign", ` from the example.
+- `packages/tasks/README.md`, lines 193-196: "A coder role's `crossRole`
+  grant of `task.reassign` can't be used. The writer check refuses it before
+  authority is looked at. …" This will be stale too. Suggested text: "The
+  coder role holds no `task.reassign` grant (row 73). The writer check
+  refuses a non-pm `task.reassign` before authority is looked at anyway. If
+  a coder should hand work on, that's a change to the field table, not to
+  the role file."
+- `packages/tasks/live/run.mjs`, line 32: the live run's own copy of the
+  slice 1 authority still gives the coder `task.reassign`. Nothing runs it in
+  the suites. It never has the coder call `task.reassign`, so its result
+  doesn't change. It should match `world.mjs`.
+- Unaffected: `scripts/test-task.sh` line 279 and
+  `business/tests/role.test.mjs` line 115 are role-validation fixtures, not
+  the coder's grant. Historical notes are left as they are: the slice 1
+  plan, the Darkwing and Filbert work notes, and lead decision 90.
+
+## What a coder's task.reassign does end to end after the change
+
+1. A pi coder session has no `task_reassign` tool, so it can't make the
+   call through the harness.
+2. It can still `raise_decision` with `action: task.reassign`, because the
+   tool's action enum is every action. The broker classes it `gated` and
+   routes it to the human, where before it went to the pm as cross-role. The
+   human is then asked a question that can only end in `field-writer`.
+3. If the human approves and the coder calls the verb directly on the bus,
+   the broker's `authorize` passes as `gated`; the bus test shows that. The
+   task verbs refuse first with `field-writer`, before `authorize`, so nothing
+   is consumed and nothing is written (tasks test).
+
+The brief names the follow-up but doesn't make it, and nor does this row. To
+refuse a non-pm `task.reassign` (or any `PM_VERBS` action) at
+`decision.raise`, before a decision opens, is a broker change and would be
+its own row.
+
+A bus that is already running keeps the authority it loaded at boot. The
+change reaches it on its next restart.
+
+## Mutants (`out/mutants-candidate.txt`, `out/mutants-base.txt`)
+
+| Mutant | Edit | Base `324c064b` | Candidate |
+|---|---|---|---|
+| `role-keeps-reassign` | `roles/coder.json` puts `task.reassign` back | n/a (the base is this mutant) | killed by role, classify, bus and harness tests |
+| `broker-unlisted-cross` | `#classification`: an action in neither list → `cross-role` | **survived** | killed by the new bus test |
+| `resolve-cross-unnarrowed` | `resolveInstance`: `crossRole` ignores `limits.authority` | killed | killed (by the narrowing test and the harness test "an action outside the instance's authority has no tool") |
+| `classify-unlisted-cross` | `classify`: an unlisted action → `cross` | killed | killed |
+| `tools-all-task-verbs` | `typedTools`: a task tool for every verb, held or not | killed | killed (now also by the new harness test) |
+
+The script saves each file, applies the sed and checks it changed something.
+It then runs the five test files and restores the file from the saved copy,
+so it works on an uncommitted tree. `out/mutants-*/` holds each diff and log.
+I ran the candidate twice. The run recorded here uses the final script; the
+first run used the same script before a comment-only reflow, with the same
+results.
+
+## Gate (`out/gate/`, sequential, scratch worktree at 324c064b with the candidate, `node_modules` linked, 23:09:50Z-23:14:34Z)
+
+| Suite | Result |
+|---|---|
+| tasks `node --test 'packages/tasks/tests/*.test.mjs'` | 71 pass, 0 fail |
+| bus | 81 pass, 0 fail |
+| business | 60 pass, 0 fail |
+| harness | 58 pass, 0 fail, 0 skipped |
+| cli | 112 pass, 0 fail |
+| webui | 50 pass, 0 fail |
+| seat | 27 pass, 0 fail |
+| test-auth | 15 passed, 0 failed |
+| test-conductor | 17 passed, 0 failed |
+| test-config | 24 passed, 0 failed |
+| test-discord | 66 passed, 0 failed |
+| test-extension-package | 18 passed, 0 failed |
+| test-foundation | 44 passed, 0 failed |
+| test-queue | 27 passed, 0 failed |
+| test-release | 14 passed, 0 failed |
+| test-task | 98 passed, 0 failed (Docker) |
+
+cli, webui and seat go beyond the gate the brief names. I ran them because
+they read the shipped roles. After the gate I removed the `rocko-r73_default`
+network.
diff --git a/agents/rocko/work/queue-73/base.txt b/agents/rocko/work/queue-73/base.txt
new file mode 100644
index 000000000..0faf48a68
--- /dev/null
+++ b/agents/rocko/work/queue-73/base.txt
@@ -0,0 +1 @@
+324c064b2a9290ff6c17449c3fa0725aa7a09494
diff --git a/agents/rocko/work/queue-73/build.patch b/agents/rocko/work/queue-73/build.patch
new file mode 100644
index 000000000..e0e721f37
--- /dev/null
+++ b/agents/rocko/work/queue-73/build.patch
@@ -0,0 +1,210 @@
+diff --git a/agents/rocko/work/queue-73/mutants.sh b/agents/rocko/work/queue-73/mutants.sh
+new file mode 100755
+index 00000000..8d550d1f
+--- /dev/null
++++ b/agents/rocko/work/queue-73/mutants.sh
+@@ -0,0 +1,56 @@
++#!/usr/bin/env bash
++# Row 73 (#1567): rerun the role-file revert and the mutants against a tree
++# with the candidate. Usage: mutants.sh <tree>. Each mutant edits one line,
++# runs the row's tests, and restores the file from a saved copy, so the tree
++# need not be committed. Prints "killed" or "SURVIVED" per mutant; exit 1 if
++# any survives, a sed doesn't change its file, or the unmutated run fails.
++# The tasks test is in the list but kills nothing here: the verb refuses a
++# coder's task.reassign with or without the grant.
++set -euo pipefail
++TREE="$(cd "${1:?usage: mutants.sh <tree>}" && pwd)"
++cd "$TREE"
++ROLE=roles/coder.json
++BROKER=packages/bus/src/broker.mjs
++RESOLVE=packages/business/src/resolve.mjs
++TOOLS=packages/harness/src/tools.mjs
++TESTS=(
++  packages/bus/tests/broker.test.mjs
++  packages/business/tests/role.test.mjs
++  packages/business/tests/resolve.test.mjs
++  packages/harness/tests/tools.test.mjs
++  packages/tasks/tests/verbs.test.mjs
++)
++run_tests() { node --test "${TESTS[@]}" >"$1" 2>&1; }
++
++# name | file | sed expression
++MUTANTS=(
++  'role-keeps-reassign|'"$ROLE"'|s/"crossRole": \["task.scope.change"\]/"crossRole": ["task.reassign", "task.scope.change"]/'
++  'broker-unlisted-cross|'"$BROKER"'|/^        ? .cross-role.$/{n;s/^        : .gated.;$/        : '"'"'cross-role'"'"';/}'
++  'resolve-cross-unnarrowed|'"$RESOLVE"'|s/^    crossRole = crossRole.filter((a) => vars\["limits.authority"\].includes(a));$/    crossRole = [...crossRole];/'
++  'classify-unlisted-cross|'"$RESOLVE"'|/^  if (resolved.limits.authority.crossRole.includes(action)) return "cross";$/{n;s/return "gated";/return "cross";/}'
++  'tools-all-task-verbs|'"$TOOLS"'|s/^      if (held.has(action)) add(name, action, action, description, obj(properties, required));$/      add(name, action, action, description, obj(properties, required));/'
++)
++
++OUT="${OUT:-$TREE/.mutants}"
++mkdir -p "$OUT"
++run_tests "$OUT/baseline.log" || { echo "baseline run failed: $OUT/baseline.log" >&2; exit 1; }
++echo "baseline: pass"
++status=0
++for m in "${MUTANTS[@]}"; do
++  IFS='|' read -r name file expr <<<"$m"
++  cp -p "$file" "$OUT/orig"
++  sed -i -e "$expr" "$file"
++  if cmp -s "$OUT/orig" "$file"; then
++    echo "$name: NOT APPLIED (sed matched nothing)"; status=1; continue
++  fi
++  diff -u "$OUT/orig" "$file" >"$OUT/$name.diff" || true
++  if run_tests "$OUT/$name.log"; then
++    echo "$name: SURVIVED"; status=1
++  else
++    echo "$name: killed ($(grep -E '^✖' "$OUT/$name.log" | grep -v 'failing tests' | sed 's/ ([0-9.]*ms)$//' | sort -u | tr '\n' ';'))"
++  fi
++  cp -p "$OUT/orig" "$file"
++  cmp -s "$OUT/orig" "$file" || { echo "restore failed: $file" >&2; exit 1; }
++  rm -f "$OUT/orig"
++done
++exit "$status"
+diff --git a/packages/bus/tests/broker.test.mjs b/packages/bus/tests/broker.test.mjs
+index 3e908d70..d5d53d9e 100644
+--- a/packages/bus/tests/broker.test.mjs
++++ b/packages/bus/tests/broker.test.mjs
+@@ -4,6 +4,9 @@ import { mkdtempSync, rmSync } from 'node:fs';
+ import { tmpdir } from 'node:os';
+ import { join } from 'node:path';
+ import { Store } from '../src/store.mjs';
++import { busBusiness } from '../src/business.mjs';
++import { resolveInstance, validateBusinessDocument } from '../../business/src/index.mjs';
++import { businessDoc, REPO_ROLES, systemFor } from '../../business/tests/helpers.mjs';
+ const { Broker } = await import('../src/broker.mjs').catch((e) => {
+   if (e.code === 'ERR_MODULE_NOT_FOUND') return {};
+   throw e;
+@@ -354,3 +357,26 @@ test('both arbiters require human resolution when their cross-role route is them
+     assert.equal(b.authorize(cap, action, { decision: d.id, target: 'vikunja:3/41' }).class, 'cross-role');
+   }
+ });
++
++test('the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role', (t) => {
++  const root = mkdtempSync(join(tmpdir(), 'bus-roles-'));
++  t.after(() => rmSync(root, { recursive: true, force: true }));
++  const business = validateBusinessDocument(businessDoc(root, 'demo'), join(root, 'businesses', 'demo.json'), {
++    rolesDir: REPO_ROLES,
++  });
++  const resolved = {};
++  for (const instance of Object.keys(business.roles))
++    resolved[instance] = resolveInstance({ system: systemFor(root), business, instance });
++  const { b, agent, human } = setup(t, { demo: busBusiness(business, resolved) });
++  const c = agent('coder');
++  call(b, c, 'role.claim');
++  const reassign = raise(b, c, 'task.reassign', { target: 'vikunja:3/7' });
++  assert.equal(reassign.class, 'gated');
++  assert.equal(reassign.route_to, 'human');
++  const scope = raise(b, c, 'task.scope.change', { domain: 'technical', target: 'vikunja:3/7' });
++  assert.equal(scope.class, 'cross-role');
++  assert.equal(scope.route_to, 'cto');
++  // Approval passes the broker; the task verb still refuses a coder with field-writer (packages/tasks).
++  call(b, human, 'decision.resolve', { id: reassign.id, choice: 'yes' });
++  assert.equal(b.authorize(c, 'task.reassign', { decision: reassign.id, target: 'vikunja:3/7' }).class, 'gated');
++});
+diff --git a/packages/business/tests/resolve.test.mjs b/packages/business/tests/resolve.test.mjs
+index 0995eb5c..75ac2925 100644
+--- a/packages/business/tests/resolve.test.mjs
++++ b/packages/business/tests/resolve.test.mjs
+@@ -122,21 +122,23 @@ test("limits.authority without role.launch leaves the launcher with no launch bl
+ });
+ 
+ test("limits.authority narrows cross-role actions too", () => {
++  // pm, the role with two cross-role actions: the limit keeps one and drops the other.
+   const s = setup((d) => {
+-    d.roles.coder.vars["limits.authority"] = ["task.update.assigned", "git.push.working", "task.scope.change"];
++    d.roles.pm.vars["limits.authority"] = ["task.create", "task.close", "task.priority.change"];
+   });
+-  const coder = resolve(s, "coder");
+-  assert.deepEqual(coder.limits.authority, { withinRole: ["task.update.assigned", "git.push.working"], crossRole: ["task.scope.change"] });
+-  assert.equal(classify(coder, "task.reassign"), "gated");
+-  assert.equal(classify(coder, "task.scope.change"), "cross");
+-  assert.equal(classify(coder, "review.request"), "gated");
++  const pm = resolve(s, "pm");
++  assert.deepEqual(pm.limits.authority, { withinRole: ["task.create", "task.close"], crossRole: ["task.priority.change"] });
++  assert.equal(classify(pm, "task.priority.change"), "cross");
++  assert.equal(classify(pm, "task.scope.change"), "gated");
++  assert.equal(classify(pm, "task.assign"), "gated");
+ });
+ 
+ test("classify", () => {
+   const s = setup();
+   const coder = resolve(s, "coder");
+   assert.equal(classify(coder, "git.push.working"), "within");
+-  assert.equal(classify(coder, "task.reassign"), "cross");
++  assert.equal(classify(coder, "task.scope.change"), "cross");
++  assert.equal(classify(coder, "task.reassign"), "gated");
+   assert.equal(classify(coder, "task.close"), "gated");
+   assert.equal(classify(coder, "deploy"), "gated");
+   assert.equal(classify(resolve(s, "pm"), "task.priority.change"), "cross");
+diff --git a/packages/business/tests/role.test.mjs b/packages/business/tests/role.test.mjs
+index 1a401307..00628f80 100644
+--- a/packages/business/tests/role.test.mjs
++++ b/packages/business/tests/role.test.mjs
+@@ -61,6 +61,8 @@ test("shipped role scopes match addendum B section 2 and the SR runbook", () =>
+ test("shipped authority follows the note's table", () => {
+   const auth = (name) => loadRole(REPO_ROLES, name).authority;
+   assert.ok(auth("coder").withinRole.includes("git.push.working"));
++  // Assignee is pm-only (decision 90): a coder grant of task.reassign could never succeed.
++  assert.deepEqual(auth("coder").crossRole, ["task.scope.change"]);
+   assert.ok(!auth("reviewer").withinRole.includes("git.push.working"));
+   assert.deepEqual(auth("reviewer").crossRole, []);
+   assert.ok(auth("pm").withinRole.includes("role.launch"));
+diff --git a/packages/harness/tests/tools.test.mjs b/packages/harness/tests/tools.test.mjs
+index b520dd4c..31606131 100644
+--- a/packages/harness/tests/tools.test.mjs
++++ b/packages/harness/tests/tools.test.mjs
+@@ -35,6 +35,14 @@ test("a coder gets no launch, no resolve_decision, and no task tools without a t
+   assert.ok(!untyped.includes("message.send"));
+ });
+ 
++test("a coder with a tracker gets its own task tools and no task_reassign", (t) => {
++  const coder = resolvedFor(t, "coder");
++  assert.deepEqual(names(typedTools({ resolved: coder, tracker: true })), [
++    "send_message", "task_scope_change", "task_update_assigned",
++    "raise_decision", "list_agents", "inbox", "trail", "list_tasks",
++  ]);
++});
++
+ test("launch only when the business's launch block names the instance as launcher", (t) => {
+   const pm = resolvedFor(t, "pm", (d) => delete d.launch);
+   assert.equal(pm.launch, null);
+diff --git a/packages/tasks/tests/verbs.test.mjs b/packages/tasks/tests/verbs.test.mjs
+index f5621f7b..c9ea98a6 100644
+--- a/packages/tasks/tests/verbs.test.mjs
++++ b/packages/tasks/tests/verbs.test.mjs
+@@ -93,6 +93,9 @@ test('assign and reassign move the role bots and record task.assigned', async (t
+   // The field-writer check runs before authority, so the refused call consumes nothing.
+   await assert.rejects(w.call(coder, 'task.assign', { task_ref, role: 'coder' }), refusedWith('field-writer'));
+   assert.equal(allowed(w, 'task.assign').length, 0);
++  // Assignee stays pm-only (decision 90); the coder role holds no task.reassign grant, and the verb refuses it anyway.
++  await assert.rejects(w.call(coder, 'task.reassign', { task_ref, role: 'reviewer' }), refusedWith('field-writer'));
++  assert.equal(allowed(w, 'task.reassign').length, 0);
+   await assert.rejects(w.call(pm, 'task.assign', { task_ref, role: 'cto' }), refusedWith('unknown-role'));
+   await assert.rejects(w.call(pm, 'task.assign', { task_ref, role: 'toString' }), refusedWith('unknown-role'));
+   await assert.rejects(w.call(pm, 'task.reassign', { task_ref, role: 'coder' }), refusedWith('task-unassigned'));
+diff --git a/packages/tasks/tests/world.mjs b/packages/tasks/tests/world.mjs
+index d50e96ea..5085e779 100644
+--- a/packages/tasks/tests/world.mjs
++++ b/packages/tasks/tests/world.mjs
+@@ -42,7 +42,7 @@ export const AUTHORITY = {
+   },
+   coder: {
+     withinRole: ['task.update.assigned', 'git.push.working', 'review.request', 'message.send'],
+-    crossRole: ['task.reassign', 'task.scope.change'],
++    crossRole: ['task.scope.change'],
+   },
+   reviewer: { withinRole: ['review.verdict', 'message.send', 'task.update.assigned'], crossRole: [] },
+   cto: {
+diff --git a/roles/coder.json b/roles/coder.json
+index 2a7325a1..04220271 100644
+--- a/roles/coder.json
++++ b/roles/coder.json
+@@ -7,7 +7,7 @@
+   "network": "api-only",
+   "authority": {
+     "withinRole": ["task.update.assigned", "git.push.working", "review.request", "message.send"],
+-    "crossRole": ["task.reassign", "task.scope.change"]
++    "crossRole": ["task.scope.change"]
+   },
+   "credentials": [
+     {
diff --git a/agents/rocko/work/queue-73/candidate-manifest.sha256 b/agents/rocko/work/queue-73/candidate-manifest.sha256
new file mode 100644
index 000000000..b8a47a7b5
--- /dev/null
+++ b/agents/rocko/work/queue-73/candidate-manifest.sha256
@@ -0,0 +1,8 @@
+e3300acd1f85f7c7c495dc2228ec904c0af992f9ea09693adcf3e5b1121d5f83  agents/rocko/work/queue-73/mutants.sh
+6c451be1461b903e2072af0a592b1be33b2e4cf9245e2fa8609ab678d7f39848  packages/business/tests/resolve.test.mjs
+17fbdd672e9daca32401b96d8cad0893aaa2ef14db9f3a6f4834e9f8b8f1b50c  packages/business/tests/role.test.mjs
+bca2e886b021831ada150dfa76be247dbe489072fb249d1a5b63e98582077853  packages/bus/tests/broker.test.mjs
+9cd45207694693734207ded38b65c99e59787db74e562668b88a499c4454ed0f  packages/harness/tests/tools.test.mjs
+5df33892007143de7537ff97059762372c78041ad3bd8be456758d3c51ec984b  packages/tasks/tests/verbs.test.mjs
+2d1c06d38617d1affd58034f5a5441fe1413b7bc2a2333fedda99c6e26baf2f0  packages/tasks/tests/world.mjs
+5b00c49dc1c998156798e0fd9fbe1e1c7e2bb62373dc9f80efaa60902b66b228  roles/coder.json
diff --git a/agents/rocko/work/queue-73/files.txt b/agents/rocko/work/queue-73/files.txt
new file mode 100644
index 000000000..0b1813aa8
--- /dev/null
+++ b/agents/rocko/work/queue-73/files.txt
@@ -0,0 +1,8 @@
+agents/rocko/work/queue-73/mutants.sh
+packages/business/tests/resolve.test.mjs
+packages/business/tests/role.test.mjs
+packages/bus/tests/broker.test.mjs
+packages/harness/tests/tools.test.mjs
+packages/tasks/tests/verbs.test.mjs
+packages/tasks/tests/world.mjs
+roles/coder.json
diff --git a/agents/rocko/work/queue-73/mutants.sh b/agents/rocko/work/queue-73/mutants.sh
new file mode 100755
index 000000000..8d550d1fb
--- /dev/null
+++ b/agents/rocko/work/queue-73/mutants.sh
@@ -0,0 +1,56 @@
+#!/usr/bin/env bash
+# Row 73 (#1567): rerun the role-file revert and the mutants against a tree
+# with the candidate. Usage: mutants.sh <tree>. Each mutant edits one line,
+# runs the row's tests, and restores the file from a saved copy, so the tree
+# need not be committed. Prints "killed" or "SURVIVED" per mutant; exit 1 if
+# any survives, a sed doesn't change its file, or the unmutated run fails.
+# The tasks test is in the list but kills nothing here: the verb refuses a
+# coder's task.reassign with or without the grant.
+set -euo pipefail
+TREE="$(cd "${1:?usage: mutants.sh <tree>}" && pwd)"
+cd "$TREE"
+ROLE=roles/coder.json
+BROKER=packages/bus/src/broker.mjs
+RESOLVE=packages/business/src/resolve.mjs
+TOOLS=packages/harness/src/tools.mjs
+TESTS=(
+  packages/bus/tests/broker.test.mjs
+  packages/business/tests/role.test.mjs
+  packages/business/tests/resolve.test.mjs
+  packages/harness/tests/tools.test.mjs
+  packages/tasks/tests/verbs.test.mjs
+)
+run_tests() { node --test "${TESTS[@]}" >"$1" 2>&1; }
+
+# name | file | sed expression
+MUTANTS=(
+  'role-keeps-reassign|'"$ROLE"'|s/"crossRole": \["task.scope.change"\]/"crossRole": ["task.reassign", "task.scope.change"]/'
+  'broker-unlisted-cross|'"$BROKER"'|/^        ? .cross-role.$/{n;s/^        : .gated.;$/        : '"'"'cross-role'"'"';/}'
+  'resolve-cross-unnarrowed|'"$RESOLVE"'|s/^    crossRole = crossRole.filter((a) => vars\["limits.authority"\].includes(a));$/    crossRole = [...crossRole];/'
+  'classify-unlisted-cross|'"$RESOLVE"'|/^  if (resolved.limits.authority.crossRole.includes(action)) return "cross";$/{n;s/return "gated";/return "cross";/}'
+  'tools-all-task-verbs|'"$TOOLS"'|s/^      if (held.has(action)) add(name, action, action, description, obj(properties, required));$/      add(name, action, action, description, obj(properties, required));/'
+)
+
+OUT="${OUT:-$TREE/.mutants}"
+mkdir -p "$OUT"
+run_tests "$OUT/baseline.log" || { echo "baseline run failed: $OUT/baseline.log" >&2; exit 1; }
+echo "baseline: pass"
+status=0
+for m in "${MUTANTS[@]}"; do
+  IFS='|' read -r name file expr <<<"$m"
+  cp -p "$file" "$OUT/orig"
+  sed -i -e "$expr" "$file"
+  if cmp -s "$OUT/orig" "$file"; then
+    echo "$name: NOT APPLIED (sed matched nothing)"; status=1; continue
+  fi
+  diff -u "$OUT/orig" "$file" >"$OUT/$name.diff" || true
+  if run_tests "$OUT/$name.log"; then
+    echo "$name: SURVIVED"; status=1
+  else
+    echo "$name: killed ($(grep -E '^✖' "$OUT/$name.log" | grep -v 'failing tests' | sed 's/ ([0-9.]*ms)$//' | sort -u | tr '\n' ';'))"
+  fi
+  cp -p "$OUT/orig" "$file"
+  cmp -s "$OUT/orig" "$file" || { echo "restore failed: $file" >&2; exit 1; }
+  rm -f "$OUT/orig"
+done
+exit "$status"
diff --git a/agents/rocko/work/queue-73/out/gate.sh b/agents/rocko/work/queue-73/out/gate.sh
new file mode 100755
index 000000000..2d83fd81e
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate.sh
@@ -0,0 +1,16 @@
+#!/usr/bin/env bash
+set -u
+W=/mnt/storage/scratch/rocko-r73; G=/mnt/storage/scratch/rocko-r73-out/gate
+mkdir -p $G; cd $W
+date -u +%Y-%m-%dT%H:%M:%SZ > $G/start.txt
+: > $G/summary.txt
+for p in tasks bus business harness cli webui seat; do
+  node --test "packages/$p/tests/*.test.mjs" > $G/$p.log 2>&1; rc=$?
+  echo "$p rc=$rc $(grep -E '^ℹ (pass|fail|skipped)' $G/$p.log | tr '\n' ' ')" >> $G/summary.txt
+done
+for s in scripts/test-*.sh; do
+  n=$(basename $s .sh)
+  bash $s > $G/$n.log 2>&1; rc=$?
+  echo "$n rc=$rc $(grep -iE '[0-9]+ passed|passed.*failed|failed' $G/$n.log | tail -1)" >> $G/summary.txt
+done
+date -u +%Y-%m-%dT%H:%M:%SZ > $G/end.txt
diff --git a/agents/rocko/work/queue-73/out/gate/bus.log b/agents/rocko/work/queue-73/out/gate/bus.log
new file mode 100644
index 000000000..727196cce
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/bus.log
@@ -0,0 +1,89 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (75.005416ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (46.010344ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (117.296262ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (88.211853ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (45.454ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (40.349948ms)
+✔ task action subjects and linked decision trail are complete and ordered (62.71095ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (19.640675ms)
+✔ business isolation includes inherited object names and cross-business message references (30.687724ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (35.606534ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (14.931591ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (30.298419ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (23.683738ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (37.164098ms)
+✔ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (36.167453ms)
+✔ S1 adapter takes resolved limits and refs, rejects mismatched instance, never mutates input (22.223275ms)
+✔ only validated broker references load; returned data and exceptions cannot expose a known token (6.415097ms)
+✔ bad file modes, symlinks, repository/data paths, malformed tokens and missing dates refuse (7.757975ms)
+✔ expiry refuses use and env references never become client data (3.599696ms)
+✔ S1 parsed service refs work, service mismatch refuses, Gitea rotation due is a warning state (3.510469ms)
+✔ opaque tokens shorter than 16 characters refuse before use (0.519448ms)
+✔ roles lists open decisions routed to a role, and never the human inbox (62.978066ms)
+✔ recent lists closed decisions of every route, newest close first, with how each closed (53.783969ms)
+✔ recent stops at RECENT rows (353.1222ms)
+✔ every row is what trail already gives the same session for that id (61.037502ms)
+✔ a role session can't list decisions, so it can't read another role's open decisions (50.943117ms)
+✔ decisions is a read: one business, strict arguments, nothing written (68.290864ms)
+✔ endLaunch writes session.ended, releases the run claim and kills its capabilities (48.870662ms)
+✔ endLaunch refuses an unknown run, a second end and a rebind of the ended run (29.248538ms)
+✔ a restarted broker refuses to rebind an ended run; a refused rebind leaves the run unbound (32.637699ms)
+✔ endLaunch leaves a claim another run took alone (151.775432ms)
+✔ refuse records action.refused against the caller with the code only (25.854569ms)
+✔ launches off refuses role.launch with launch-revoked until launches on (41.369233ms)
+✔ broker process: launch ops authorize role.launch, record refusals and end runs (183.620622ms)
+✔ human proof binds CLI entry, process start and nonce; agents and incomplete ancestry refuse (5.460808ms)
+✔ process reader gets own kernel identity without exposing environment values (1.105513ms)
+✔ EACCES ancestor environments skip only markers; commands and registered launches still refuse (2.890964ms)
+✔ real pid 1 remains inspectable when its environment is protected (0.66196ms)
+✔ within-role sends cite an open gated launch decision without spending it or naming it in grants (58.516738ms)
+✔ missing and foreign-business citations refuse and roll back message and grant (33.696575ms)
+✔ cross-role sends still need a matching resolved decision and consume it once (51.512186ms)
+✔ broker process binds trusted launches, offers reader capabilities, refuses human mutation, closes cleanly (140.146767ms)
+✔ startup token refusal returns safe code without value or partial listening broker (165.369856ms)
+✔ loaded fixture token is absent from socket replies and SQLite, including refusal evidence (138.944712ms)
+✔ killed broker leaves an explicit stale lock; another process cannot silently reclaim it (186.620794ms)
+✔ trusted host registers later launches; socket clients never have a registration verb (143.38202ms)
+✔ runtime excludes declared project roots even when host supplies no repoRoots (181.87432ms)
+✔ a refused launch binding leaves the broker and existing capabilities alive; bad protocol stops it (81.246986ms)
+✔ v3b prototype refusals, views and append-only mutations (169.826249ms)
+✔ gated approval authorizes once, survives store reopen, and fresh approval works (66.31408ms)
+✔ another run cannot consume an approval; a failed check leaves it usable (170.806451ms)
+✔ two scheduled callers have exactly one grant and one consumed refusal (35.10518ms)
+✔ failed commit rolls consumption back; cross-role consumes and within-role stays reusable (71.095234ms)
+✔ class drift gated to cross-role refuses before consumption (41.943308ms)
+✔ class drift cross-role to gated refuses before consumption (42.424786ms)
+✔ class drift gated to within-role refuses before consumption (30.638739ms)
+✔ class drift cross-role to within-role refuses before consumption (32.766764ms)
+✔ class drift within-role to gated refuses before consumption (36.477882ms)
+✔ class drift within-role to cross-role refuses before consumption (41.694199ms)
+✔ message.send consumes approval and prevents a later send or authorize (36.671986ms)
+✔ role.revoke consumes approval and prevents a later revoke or authorize (38.981706ms)
+✔ creates private WAL store and excludes a second writer until explicit close (32.015978ms)
+✔ rollback is atomic and schema metadata is checked against trusted DDL, not just itself (31.440407ms)
+✔ existing empty database and symlink runtime directory refuse, never initialize over damage (30.603692ms)
+✔ crash during a transaction recovers no partial event after explicit fixture-only lock removal (197.089583ms)
+✔ writer refuses mixed at/read_at forms atomically, even through trusted SQL helpers (24.893809ms)
+✔ async transactions refuse before invoking their function (24.126915ms)
+✔ recordTask keeps sync reads and a role write apart (50.271252ms)
+✔ read_at must be one canonical UTC format, so the projection compares strings safely (18.59157ms)
+✔ a bad entry refuses the whole record (28.135164ms)
+✔ taskView reads the projection for one business (155.463221ms)
+✔ requestTask hands only a holder and a task verb to the handler, and records refusals (64.058729ms)
+✔ the server sends task verbs to the adapter with its own timeout; other verbs stay synchronous (308.778218ms)
+✔ without an adapter the server refuses every task verb (42.008397ms)
+✔ the runtime refuses an invalid adapter and closes a valid one (52.987183ms)
+✔ the process loads the S3 adapter from plain-data trackers (288.225403ms)
+✔ socket capability stamps launch identity; shared views use wire, no SQL client (51.411447ms)
+✔ two wire claims serialize; a lost reply never automatically retries (185.785126ms)
+✔ malformed, oversized and identity-forging envelopes refuse without echoing input (33.328628ms)
+✔ client preserves UTF-8 when a response divides a multibyte character (23.91873ms)
+✔ committed mutation followed by dropped reply reports unknown and is never retried (27.83698ms)
+ℹ tests 81
+ℹ suites 0
+ℹ pass 81
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1202.758647
diff --git a/agents/rocko/work/queue-73/out/gate/business.log b/agents/rocko/work/queue-73/out/gate/business.log
new file mode 100644
index 000000000..075582a14
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/business.log
@@ -0,0 +1,68 @@
+✔ config directory and file path follow MOSAIC_CONFIG (3.499142ms)
+✔ the fixture business validates and comes back frozen (8.200955ms)
+✔ two instances may share a definition (2.889005ms)
+✔ top-level refusals (9.161253ms)
+✔ arbiters and projects (18.302779ms)
+✔ role instances (7.565324ms)
+✔ Vikunja bots (14.375041ms)
+✔ a role without Vikunja takes no tracker block (7.070399ms)
+✔ credential references match the definition's services (8.726652ms)
+✔ launch (19.472588ms)
+✔ loadBusiness: file checks (4.217411ms)
+✔ loadBusiness: not a regular file (75.414373ms)
+✔ loading writes nothing (3.026667ms)
+✔ names that are Object.prototype properties don't count as declared (6.226665ms)
+✔ the shipped example refuses as written and validates once filled in (1.219599ms)
+✔ usage errors exit 4 (477.425638ms)
+✔ validate: a good business exits 0 and prints instance digests (122.226689ms)
+✔ validate: project files (587.623581ms)
+✔ validate: missing files and a broken system config (412.505364ms)
+✔ validate: credential reference problems exit 2 and name each one (114.443595ms)
+✔ validate: a token file inside the repository is refused (115.872029ms)
+✔ validate: role definitions come from MOSAIC_ROLES_DIR (395.576524ms)
+✔ resolve: prints one instance's record (382.072144ms)
+✔ resolve: refusals (671.297775ms)
+✔ parse: exactly one of file or env, plus the service's date (5.20369ms)
+✔ check: a good file has no problems (3.750629ms)
+✔ check never opens the file: a write-only token passes (0.883724ms)
+✔ check: file problems (2.127762ms)
+✔ check: token files can't live in the repository or dataRoot, even through a linked directory (1.608643ms)
+✔ check: dates and environment references (1.54162ms)
+✔ path and load (3.887149ms)
+✔ refusals (3.166373ms)
+✔ systemVars flattens the validated config (3.977377ms)
+✔ precedence: system, business, project, project role, agent (9.369246ms)
+✔ limits narrow the definition and never widen it (5.814447ms)
+✔ role.launch stays within-role only for the instance the launch block names (9.938752ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (3.698816ms)
+✔ limits.authority narrows cross-role actions too (3.358156ms)
+✔ classify (2.651364ms)
+✔ the record carries what the broker and launcher need (1.878428ms)
+✔ digest: key order doesn't matter, any value change does (16.959853ms)
+✔ refusals (5.200785ms)
+✔ the four shipped version 2 roles load (6.343277ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.898026ms)
+✔ shipped authority follows the note's table (0.923704ms)
+✔ version 1 files keep loading with no authority (1.694498ms)
+✔ the conductor policy isn't a role (0.391224ms)
+✔ a missing role file is exit 4, a symbolic link too (1.024239ms)
+✔ version 2 refusals (2.475814ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (4.492413ms)
+✔ credentials: Gitea scopes (2.331614ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (4.431165ms)
+✔ credentials: services (1.431893ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.71708ms)
+✔ every key names known layers and a merge rule (1.619414ms)
+✔ unknown keys and wrong layers refuse (1.319144ms)
+✔ types (3.666892ms)
+✔ merge: defaults, then the most specific layer wins (0.481573ms)
+✔ merge: limits only narrow, and provenance lists each source (0.550264ms)
+✔ merge doesn't change its inputs (0.231736ms)
+ℹ tests 60
+ℹ suites 0
+ℹ pass 60
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 3407.146697
diff --git a/agents/rocko/work/queue-73/out/gate/cli.log b/agents/rocko/work/queue-73/out/gate/cli.log
new file mode 100644
index 000000000..a04778754
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/cli.log
@@ -0,0 +1,122 @@
+✔ every screen and overlay fits 120x36, 80x24 and the narrow layout, row for row (2575.437868ms)
+✔ width counts terminal cells: wide CJK and emoji 2, U+FE0F after a one-cell base 1, marks 0 (121.840489ms)
+✔ broker text is shown, never interpreted: no raw control character or bidi override reaches the terminal (259.988312ms)
+✔ NO_COLOR: no escape codes, except reverse video on the selected row (583.254542ms)
+✔ 1-9 shows the decide flow's own lines and question; only y resolves (75.499086ms)
+✔ the app shows the same decide text as mosaic decide (31.276476ms)
+✔ any key but y or Y at the question answers No; nothing is sent (246.91868ms)
+✔ below 40x10 the question is hidden, so every key answers it No, y included (Filbert R3) (82.445862ms)
+✔ a throw while answering the question goes to crash, sends nothing and leaves no job pending (25.837195ms)
+✔ type-ahead never answers the question: a paste, keys typed before it is drawn, keys inside askMs (445.758625ms)
+✔ quitting at the question answers No and sends nothing, armed or not (48.00947ms)
+✔ a timer tick reads nothing under an overlay, never answers the question, and keeps a resolved status (37.142973ms)
+✔ a long question is cut inside the question; the action, target, choice and prompt stay (Dewey B1) (46.730692ms)
+✔ a decide box too big for the terminal says so, and every key answers it No, y included (Dewey B1, round 3) (367.441442ms)
+✔ at every size the decide box shows each line y depends on, or says it is hidden (13172.79353ms)
+✔ outcome unknown: the CLI's sentence, no retry, then the trail or a re-read (81.404171ms)
+✔ refused and failed reads keep the last list, marked stale; a refused decide says refused (37.386594ms)
+✔ keys: views, moving, trail of a selection, the command line, help and quit (90.340526ms)
+✔ one option: the status line says Press 1, not 1-1 (24.889131ms)
+✔ no business: the app says so and :business picks one (15.041507ms)
+✔ mosaic with no argument: exit 3 in an agent run, exit 4 with the usage off a terminal, the app on one (5.682261ms)
+✔ scripts/mosaic with no argument reaches the app entry; the seat usage it printed before has not drifted (230.572504ms)
+✔ strings to keep exactly live once, in the decide flow and the transport; the app does not restate them (0.620902ms)
+✔ decide with --note still asks; a No sends nothing (11.58382ms)
+✔ the app's business comes from a live bus host only (2.054439ms)
+✔ runApp leaves the terminal as it found it on q, end of input, SIGTERM, SIGINT, SIGHUP and an error, the first frame's included (10.430805ms)
+✔ draw is pure: same state, same frame (1.022236ms)
+✔ inbox lists only decisions routed to the human, with what approving authorizes and how to decide (54.444537ms)
+✔ decide resolves by id prefix with --yes, and the trail shows it in broker order with the task to follow (50.898151ms)
+✔ decide refuses without a terminal or --yes, on an unknown option and on a short reference (39.078896ms)
+✔ decide prints a declining choice as declining (30.366338ms)
+✔ an unknown outcome is reported once and never resent (50.934024ms)
+✔ a decision closed before the answer arrives exits 2 and points at its trail (25.95427ms)
+✔ a prefix that matches two open decisions exits 2 and resolves neither (35.958621ms)
+✔ without --business a command uses the live host's business, and a stale host.json is not a host (16.387693ms)
+✔ every human command refuses inside an agent run before it touches the bus (26.398696ms)
+✔ usage errors exit 4; no business and no host is a usage error (22.963519ms)
+✔ agents and tasks print through the broker (21.444282ms)
+✔ notify.json: missing, loose, malformed and extra keys refuse; a binding or null passes (3.686442ms)
+✔ bootConfig builds the broker's boot message for one business, with no trackers key when no project names one (126.715818ms)
+✔ trackers come from the tracker.* variables of the one project that names a tracker project (178.667084ms)
+✔ with two projects, the one that sets tracker.project supplies the tracker and the other is no conflict (76.289126ms)
+✔ two projects that each name a tracker project refuse, since the boot shape holds one (103.153178ms)
+✔ a business without tracker.baseUrl gets no trackers entry (93.104531ms)
+✔ an unknown business and a broken system config refuse with exit 3 (220.550038ms)
+✔ empty views say so (1.445675ms)
+✔ the trail keeps the broker's order and names a decision's task without its rows (1.62635ms)
+✔ tasks print the tracker fields the snapshot carries (0.275421ms)
+✔ the host boots the broker, binds a launch in process, and the notifier DMs a blocking decision exactly once (1234.573704ms)
+✔ a notifier that dies takes the host down with exit 1, so the unit restarts the pair (255.78367ms)
+✔ a second host for the same data root refuses with exit 3 while the first runs (125.006311ms)
+✔ a broker reply that misses the wait breaks the channel: the late reply answers nothing, later requests refuse, the host exits 1 (1130.72943ms)
+✔ a broker reply with another request's id, or none, breaks the channel and the host exits 1 (307.7752ms)
+✔ a broker reply with no request waiting breaks the channel and the host exits 1 (200.138306ms)
+✔ a notifier that refuses stops the broker and the host refuses with exit 3 (265.698781ms)
+✔ a notifier that refuses after the broker died still refuses with exit 3, without a send to the dead broker (181.266956ms)
+✔ a notifier that dies before it replies, after the broker died, still refuses, without a send to the dead broker (146.196068ms)
+✔ a close send that fails with EPIPE after the notifier refuses still gives the notifier's refusal, exit 3 (194.368599ms)
+✔ a close send that fails with EPIPE after the notifier dies unanswered still gives the notifier's error (114.904121ms)
+✔ close() whose stop and close sends fail with EPIPE still finishes, with exit 1 (216.017254ms)
+✔ watchChildren reports a child that died before it was called, and one that dies later (34.650091ms)
+✔ bus stop refuses to signal a live pid that is not a bus host (204.737058ms)
+✔ bus start refuses with exit 3 and the code when the broker refuses to boot; bus status names the lock (402.133502ms)
+✔ bus start refuses with exit 3 without a notifier config (193.518643ms)
+✔ bus start runs until bus stop; status reports it while it runs (938.99266ms)
+✔ bus-service.sh renders the unit and installs it into a given directory (78.270678ms)
+✔ bus-service.sh pins the claude binary, refuses one the unit can't carry, and renders without one (121.526952ms)
+✔ bus start --pm: the PM runs under its own PID namespace, registered, with a manifest (1386.034158ms)
+✔ the PM launches a coder through its launch tool; the coder answers; refusals name their code (583.053034ms)
+✔ a runner that stops at once ends its launch with the runner's reason (276.596415ms)
+✔ a host that died hard leaves its sessions to the next host, which kills them and ends their runs so the role can be launched again (902.684414ms)
+✔ a host that died hard leaves its sessions to the next host, which finds one already exited (23) and ends their runs so the role can be launched again (3857.50546ms)
+✔ a malformed sessions.json refuses the host start with exit 3 and stays as it was (110.075476ms)
+✔ an over-long launch request is refused at once, not at the 10 s idle timeout (123.391452ms)
+✔ a launch client that never closes its side doesn't hold the host's close (125.094283ms)
+✔ a runner that ignores SIGTERM is killed when the host closes (1230.598399ms)
+✔ claudeVersion asks MOSAIC_CLAUDE_BIN when set, else claude on PATH (7.358594ms)
+✔ zoned uses the IANA zone across DST (29.221001ms)
+✔ each open blocking decision is DM'd once, across polls and a restart; non-blocking ones are not (41.382705ms)
+✔ two blocking decisions get two DMs with different nonces (41.511473ms)
+✔ the digest nonce differs per business and per day and fits Discord's 25 characters (0.391142ms)
+✔ a failed DM is journaled, backs off, and is retried until it lands (41.061189ms)
+✔ five definite refusals stop a DM: one gave-up line, one log line, and a restart keeps the count (56.151739ms)
+✔ 429s, 5xx-style unknowns and refusals without a status never count toward the limit (83.90553ms)
+✔ a crash between the fifth refusal and its gave-up line: the next poll appends it and sends nothing (37.52893ms)
+✔ polled every POLL_MS against a permanent 403, a DM is sent at 0, 30, 60, 90 and 120 min and gives up only then (232.090057ms)
+✔ a restart after the second refusal does not send before that refusal's 30 min are up (113.374042ms)
+✔ the digest goes at 08:00 Chicago once a day, with blocking ones marked as DM'd (36.376926ms)
+✔ a late start with no digest for the day sends one at once; an empty inbox gets one line (39.173684ms)
+✔ an inbox read failure is logged and the next poll retries (2.215435ms)
+✔ no Discord id reaches the journal or the log (22.25059ms)
+✔ the journal: a torn tail is copied out and truncated, so an append after it reopens cleanly (12.880256ms)
+✔ the journal: a crash between the copy and the truncate leaves a tail the next open repairs (6.917527ms)
+✔ the journal: a whole file that is one torn line truncates to empty (2.277526ms)
+✔ the journal: a malformed complete line refuses and leaves the file and any torn tail alone (1.171449ms)
+✔ the journal: a loose file mode, a loose directory or a symlinked journal refuses (3.129807ms)
+✔ the journal: a line with a wrong type refuses with exit 3 and names the field (8.844715ms)
+✔ the journal: a symlinked directory refuses and says it is a link (1.039649ms)
+✔ the journal: a dangling directory link, a parent that is a file and a journal that is a directory each refuse with exit 3 (1.847472ms)
+✔ the journal: an append after the file was swapped for a symlink refuses and writes nothing through it (1.433451ms)
+✔ the journal: a directory it cannot write or create refuses with exit 3 and names the path (1.114175ms)
+✔ digest content stays within Discord's 2000 characters (0.621691ms)
+✔ runLoop never overlaps ticks and stops after the one in flight (119.237271ms)
+task.close {} answered: invalid-request; fake saw 18 requests, first GET /info 200, GET /projects/1 200, GET /projects/1/views 200
+task.close on a missing task answered: task-not-found; it made GET /tasks/999 404
+✔ bootConfig trackers reach the S3 adapter in the real broker child, which goes ready against a fake Vikunja (724.50701ms)
+✔ the transport writes {business, verb, args} to the child and reads its JSON (91.241323ms)
+✔ a bus code on stderr becomes the exit code; garbage and timeouts are outcome problems (2435.284197ms)
+✔ busExit and refuseInsideAgent (0.695552ms)
+✔ talk sends a REQUEST, prints and reads everything that arrives, and stops at the reply (115.335304ms)
+✔ talk --wait 0 only sends; no reply within --wait exits 1 and says where it will show (1053.430583ms)
+✔ talk, stop and launches off/on refuse inside an agent run; bad arguments are usage errors (15.844796ms)
+✔ launches off and on go to the broker and change the business's launch state (14.087384ms)
+✔ launches list reads sessions.json and marks a stale entry; stop reports it and refuses a non-runner (15.217499ms)
+ℹ tests 112
+ℹ suites 0
+ℹ pass 112
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 18837.646924
diff --git a/agents/rocko/work/queue-73/out/gate/end.txt b/agents/rocko/work/queue-73/out/gate/end.txt
new file mode 100644
index 000000000..28d6925b7
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/end.txt
@@ -0,0 +1 @@
+2026-10-11T23:14:34Z
diff --git a/agents/rocko/work/queue-73/out/gate/harness.log b/agents/rocko/work/queue-73/out/gate/harness.log
new file mode 100644
index 000000000..c47d4b0f8
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/harness.log
@@ -0,0 +1,66 @@
+✔ sessionModel: agent vars win, then the system's execution settings (28.582565ms)
+✔ a pi bundle: prompt, policy, tools and manifest, 0600 in a 0700 directory (10.80112ms)
+✔ a claude-code bundle adds the wrapped gate hook and the MCP config (7.891606ms)
+✔ a bundle is written once: an existing file refuses (4.681172ms)
+✔ a path with a single quote can't go into the hook command (5.835675ms)
+✔ allow exits 0, a deny exits 2 with the reason on stderr (491.930816ms)
+✔ a missing or wrong policy, or a bad event, exits 2 (280.01237ms)
+✔ the bundle's wrapped command: a missing gate or node still blocks (1240.222858ms)
+✔ claude: typed tools through MCP, the hook blocks, builtins outside --tools don't exist (1816.377717ms)
+✔ claude: the hook alone blocks a path outside the workspace (731.111927ms)
+✔ claude: a second turn resumes the first turn's session (1803.113028ms)
+✔ claude adapter: --restricted is always passed (9.598654ms)
+✔ claude adapter: MOSAIC_CLAUDE_BIN wins over PATH and must be an absolute executable (43.271331ms)
+✔ claude: CLAUDE.md files and auto-memory don't reach the model; without --restricted they do (1233.642263ms)
+✔ claude: a missing hook or MCP file refuses before claude starts (13.803182ms)
+✔ pi: policy tools and typed tools pass, anything else is blocked (9.070302ms)
+✔ claude: builtins map from pi names, typed tools need the mcp prefix (1.487225ms)
+✔ file tool paths must resolve inside the workspace (7.762967ms)
+✔ pi's own path normalisation can't be used to step out (5.452659ms)
+✔ a symlink inside the workspace that points out is outside (2.168089ms)
+✔ a dangling symlink is refused at any depth, in both harnesses (8.67524ms)
+✔ read is checked under every spelling pi's read would open, in both harnesses (45.396712ms)
+✔ other spellings cover directories, dangling links and pi's cwd (3.008336ms)
+✔ a relative path climbs from the workspace's real path, in both harnesses (10.755171ms)
+✔ claude path fields per tool (1.281399ms)
+✔ glob patterns stay inside the workspace (1.64637ms)
+✔ a path that can't be checked is blocked (0.661031ms)
+✔ initialize, ping and tools/list (100.93379ms)
+✔ tools/call goes through the tool socket; a refusal is an isError result (79.489337ms)
+✔ unknown tools and methods are JSON-RPC errors and never reach the socket (73.485687ms)
+✔ a missing argument is a usage error (131.29632ms)
+✔ pi: typed tools reach the socket, the gate blocks, agent_end writes the marker (1126.108528ms)
+✔ pi: a write through a dangling symlink is blocked, and nothing appears outside (674.432288ms)
+✔ pi: a read is refused when pi would open another spelling outside (584.730201ms)
+✔ pi: a relative path climbs from the real path of a workspace behind a symlink (675.898409ms)
+✔ pi: a relative path climbs from the real path of a dataRoot behind a symlink (959.55437ms)
+✔ pi: a missing extension refuses before any model call (14.697573ms)
+✔ pi: an extension without its configuration fails pi's start (532.20757ms)
+✔ founderCheck: founder variables, then a needed service without a usable token (5.252564ms)
+✔ turnRequest names the sender, class, reply and decision (0.404605ms)
+✔ a message becomes a turn, the answer goes back as a RESULT, SIGTERM releases and exits 0 (437.11389ms)
+✔ a SIGTERM before the claim stops the runner with exit 0 and no claim (195.724419ms)
+✔ typed tools carry the runner's capability; launch goes to the host's launch socket (575.474458ms)
+✔ a RESULT gets no automatic reply; failed turns reply with the reason (1375.381179ms)
+✔ SIGTERM during a turn kills the turn's process group and still exits 0 (194.928324ms)
+✔ founder credentials stop before the claim (20) (184.605222ms)
+✔ a refused claim exits 21; an ended run's capability exits 22 (453.390314ms)
+✔ the launch ending under a running session exits 22 (217.268639ms)
+✔ a broker that stays unreachable exits 23 after brokerRetries polls (272.988015ms)
+✔ a broker that is down at the claim exits 23, not 21 (96.658628ms)
+✔ no capability, or a malformed one, on stdin exits 2 (289.734618ms)
+✔ a missing or malformed policy exits 2 before the claim (164.918924ms)
+✔ the PM gets launch, its task verbs and the reads (11.769269ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (4.543725ms)
+✔ a coder with a tracker gets its own task tools and no task_reassign (2.885134ms)
+✔ launch only when the business's launch block names the instance as launcher (2.70311ms)
+✔ an action outside the instance's authority has no tool (8.068441ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (24.367924ms)
+ℹ tests 58
+ℹ suites 0
+ℹ pass 58
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 11062.352505
diff --git a/agents/rocko/work/queue-73/out/gate/seat.log b/agents/rocko/work/queue-73/out/gate/seat.log
new file mode 100644
index 000000000..e2df11356
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/seat.log
@@ -0,0 +1,35 @@
+✔ resolveSeat: by name under --repo resolves the repo layout (2.571856ms)
+✔ resolveSeat: by path resolves the fleet layout (0.733308ms)
+✔ resolveSeat: refusals for missing dir, missing launch.sh, non-executable launch.sh, invalid name, and unknown layout (1.502708ms)
+✔ tmuxContext: outside tmux, default socket, custom socket, and exec failure (1.320435ms)
+✔ makeRegistration produces a record that validates; each shape violation throws SeatError (1.629035ms)
+✔ writeRegistration/readRegistration: round trip, permissions, absence, and malformed records (2.372825ms)
+✔ updateTask: changes task, taskSetBy and updatedAt only, and refuses appropriately (1.798218ms)
+✔ CLI launch: registers, execs the fake launch script, and passes args through (64.567607ms)
+✔ CLI launch: --harness lands in the record (57.869902ms)
+✔ CLI launch: the launch script's own exit code passes through (43.095625ms)
+✔ CLI launch: relaunching a seat rewrites the one registration record (93.424815ms)
+✔ CLI launch: omitting --task records an empty string, not null (46.406032ms)
+✔ CLI seat task: updates only the task after a launch, and refuses on an unlaunched seat (133.037855ms)
+✔ CLI refusals: no args, unknown flag, missing config, already-registered env, and exec failure (204.105953ms)
+✔ samePath: equal paths, symlinked dirs, distinct dirs, and non-strings (0.619616ms)
+✔ resolveSetBy: explicit --by wins over the environment; absent or empty environment gives unknown; invalid explicit or environment values refuse with exit 4 (0.743583ms)
+✔ validateRegistration/readRegistration: taskSetBy is optional; a record without it (written before #1511) still loads unchanged; an invalid one is refused; the version does not change (1.034ms)
+✔ updateTask: records setBy, preserves startedAt and every unrelated field, upgrades an old record in place only when the task is set, and replaces a previous attribution (1.309264ms)
+✔ CLI seat task: --by beats MOSAIC_AGENT_NAME, the environment beats nothing, empty environment is unknown, invalid --by or environment refuses with exit 4 and leaves the record byte for byte (424.850272ms)
+✔ family: exactly one launch.max key in the model name, else null (1.134819ms)
+✔ sessionEnv passes only the allowlist, the repo's bin on PATH, and the run id (1.321324ms)
+✔ newRun: short ids, 0700 directories, and a refusal when the socket path won't fit (2.696089ms)
+✔ session file and launch log: 0600, the session file written once (1.648677ms)
+✔ endReason maps the runner's exit codes; a signal is killed (0.21076ms)
+✔ Registry mirrors to sessions.json; readSessions marks live entries; bad files refuse (1.820038ms)
+✔ stopSession refuses an unknown run, reports a stale one, and won't signal a pid that isn't a runner (1.401457ms)
+✔ a session runs under unshare as pid 1 of its namespace, claims, answers, and stops on mosaic stop (250.310689ms)
+ℹ tests 27
+ℹ suites 0
+ℹ pass 27
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1192.412558
diff --git a/agents/rocko/work/queue-73/out/gate/start.txt b/agents/rocko/work/queue-73/out/gate/start.txt
new file mode 100644
index 000000000..3825455e9
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/start.txt
@@ -0,0 +1 @@
+2026-10-11T23:09:50Z
diff --git a/agents/rocko/work/queue-73/out/gate/summary.txt b/agents/rocko/work/queue-73/out/gate/summary.txt
new file mode 100644
index 000000000..2401ad323
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/summary.txt
@@ -0,0 +1,16 @@
+tasks rc=0 ℹ pass 71 ℹ fail 0 ℹ skipped 0 
+bus rc=0 ℹ pass 81 ℹ fail 0 ℹ skipped 0 
+business rc=0 ℹ pass 60 ℹ fail 0 ℹ skipped 0 
+harness rc=0 ℹ pass 58 ℹ fail 0 ℹ skipped 0 
+cli rc=0 ℹ pass 112 ℹ fail 0 ℹ skipped 0 
+webui rc=0 ℹ pass 50 ℹ fail 0 ℹ skipped 0 
+seat rc=0 ℹ pass 27 ℹ fail 0 ℹ skipped 0 
+test-auth rc=0 selftest: 15 passed, 0 failed
+test-conductor rc=0 selftest: 17 passed, 0 failed
+test-config rc=0 selftest: 24 passed, 0 failed
+test-discord rc=0 discord suite: 66 passed, 0 failed
+test-extension-package rc=0 extension package selftest: 18 passed, 0 failed
+test-foundation rc=0 selftest: 44 passed, 0 failed
+test-queue rc=0 queue suite: 27 passed, 0 failed
+test-release rc=0 selftest: 14 passed, 0 failed
+test-task rc=0 selftest: 98 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/tasks.log b/agents/rocko/work/queue-73/out/gate/tasks.log
new file mode 100644
index 000000000..c713e892b
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/tasks.log
@@ -0,0 +1,79 @@
+✔ the boot config is checked before anything starts (47.200233ms)
+✔ a business with no tracker entry refuses task verbs (26.054232ms)
+✔ credential.expiring and .expired are recorded once per instance (71.464071ms)
+✔ a token file that changes on disk records credential.changed (19.23182ms)
+✔ autostart polls, reconciles and retries a startup the tracker was down for (55.654321ms)
+✔ a refusal a restart must clear is not retried by the poll (19.019302ms)
+✔ a poll that fires while two are queued is dropped (31.522387ms)
+✔ close waits for a running verb and refuses one that has not started (84.044017ms)
+✔ the bundled Vikunja is the pinned upstream image the runbook names (1.460154ms)
+✔ every published port is on 127.0.0.1, and no secret is in the file (0.553947ms)
+✔ the fake answers each route with the statuses and shapes Vikunja v2.7.0 sent (224.349567ms)
+✔ the project list holds recorded project objects in the recorded list envelope (22.371425ms)
+✔ the recorded task bodies pass the checks S3 applies to every read (1.419284ms)
+✔ the client works against the fake over real HTTP with the platform fetch (215.956568ms)
+✔ a correct install starts, and the first reconcile records tasks that already exist (56.215912ms)
+✔ verbs refuse while a business is starting and after startup refused it (41.960644ms)
+✔ startup refuses a token that can do more than its role needs (69.306729ms)
+✔ startup refuses an unsupported version and flags an untested one (63.000628ms)
+✔ startup refuses a board that the runbook did not install (85.304122ms)
+✔ startup refuses a project the sync bot cannot read (46.47812ms)
+✔ startup refuses a configured label the pm bot cannot see (25.371916ms)
+✔ startup refuses an expired credential and a missing sync credential (52.17586ms)
+✔ an unreachable tracker refuses with tracker-unavailable (15.114537ms)
+✔ an edit in the UI is recorded once, with the fields that changed (127.212799ms)
+✔ a move between open buckets is seen on the board, though updated does not change (65.945293ms)
+✔ a person's comment is counted and a bot's is not (118.78426ms)
+✔ the hourly reconcile catches a comment through comment_count (159.932149ms)
+✔ a task closed in the UI leaves the open view with its done bucket (177.635334ms)
+✔ a task that leaves the board is recorded as deleted, moved or out of reach (196.943355ms)
+✔ a poll that read before a verb wrote does not overwrite the verb (89.064442ms)
+✔ a tracker fault during a tick is reported and the next tick catches up (53.781773ms)
+✔ a malformed answer refuses the tick with tracker-shape (22.569344ms)
+✔ no token value reaches the database, the log or a refusal (71.43717ms)
+✔ the first look at a task counts only comments inside the window (47.402383ms)
+✔ with descendants off the adapter reads no project list and no child, as in v0 (75.132178ms)
+✔ a three-level tree is read per project, and a project outside it is not (43.731266ms)
+✔ a watch-only child keeps the business ready, and a failed child read refuses as transient (52.398292ms)
+✔ a root without the installed buckets refuses with descendants on (20.347468ms)
+✔ the project list refuses without read_all, and skips archived projects and saved filters (47.255975ms)
+✔ descendants must be true or false (15.341385ms)
+✔ a move inside the tree is a moved tombstone plus a snapshot under the new ref (111.510521ms)
+✔ a child made after startup joins the tree at the next reconcile (86.324892ms)
+✔ a child that leaves the tree has its tasks recorded missing, and the ledger keeps their history (94.225955ms)
+✔ a child deleted or hidden between reconciles drops out at the next tick, and the root still refuses (146.919538ms)
+✔ a reconcile that fails keeps the previous tree (80.42665ms)
+✔ a closed task the cursor missed takes its own project's done bucket (61.092169ms)
+✔ closed tasks read at startup take their own project's done bucket (38.269422ms)
+✔ a closed task the cursor returns takes its own project's done bucket (70.162556ms)
+✔ a task moved out of a child that leaves the tree at the same time is moved, not no-access (46.700813ms)
+✔ a task in a bucket outside the five keeps its raw bucket id (38.564246ms)
+✔ bucket 0 on a watch-only task never becomes a bucket or a state (95.280726ms)
+✔ a child whose kanban view is deleted still refuses the tick, and the reconcile reads it watch-only (70.345835ms)
+✔ a child dropped at a tick comes back at the next reconcile once the bot reads it again (50.218065ms)
+✔ task.create needs a recorded human request and a requirement id (95.495374ms)
+✔ only labels named in the business file can be written (54.586778ms)
+✔ task.schedule sets and clears a due date and relations (87.727746ms)
+✔ assign and reassign move the role bots and record task.assigned (136.022256ms)
+✔ task.update.assigned is for the assignee and records task.state (158.096097ms)
+✔ a wrong expected digest records task.conflict and writes nothing (58.524297ms)
+✔ a cross-role verb needs a resolved decision, used once (75.182705ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (157.320621ms)
+✔ a lost answer is settled by a re-read and never retried (77.393226ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (77.31654ms)
+✔ a task the sync bot cannot read refuses and records nothing (36.749543ms)
+✔ verbs and polls for one business run one at a time (113.82165ms)
+✔ a due date with milliseconds is written to the second (80.352537ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (47.226617ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (27.798719ms)
+✔ a create whose final read fails succeeds and records task.created (44.517445ms)
+✔ an edit between the last write and the final read shows as external on the next poll (70.754281ms)
+✔ task.created is recorded when a later label write fails (24.516626ms)
+ℹ tests 71
+ℹ suites 0
+ℹ pass 71
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1636.446672
diff --git a/agents/rocko/work/queue-73/out/gate/test-auth.log b/agents/rocko/work/queue-73/out/gate/test-auth.log
new file mode 100644
index 000000000..ac2287606
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-auth.log
@@ -0,0 +1,17 @@
+OK    status with missing harness credential exits 3 and still lists accounts
+OK    status reports harness credential (read-only) + mosaic accounts
+OK    api key material never reaches output
+OK    oauth token material never reaches output
+OK    unparseable credential file exits 2
+OK    symlinked credential file exits 4
+OK    env-side credential names reported
+OK    env var values never reach output
+OK    accounts without an accounts dir reports none and creates nothing
+OK    accounts lists files and marks the active one
+OK    loose account perms flagged in listing
+OK    agent --auth with missing account file refuses (exit 4)
+OK    agent --auth with non-0600 account file refuses
+OK    agent --auth with invalid account name refuses
+OK    auth.sh without valid config refuses
+
+selftest: 15 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-conductor.log b/agents/rocko/work/queue-73/out/gate/test-conductor.log
new file mode 100644
index 000000000..62afc513c
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-conductor.log
@@ -0,0 +1,23 @@
+On branch refactor
+Your branch is up to date with 'origin/refactor'.
+
+nothing to commit, working tree clean
+OK    dry-run: allowed change, exit 0, nothing committed (exit 0)
+OK    dry-run committed nothing
+OK    apply: allowed change exits 0 (exit 0)
+OK    apply: attribution in commit subject
+OK    apply: target tree clean after commit
+OK    disallowed path refused (exit 1)
+OK    disallowed path: target untouched
+OK    syntax gate refused broken .mjs (exit 1)
+OK    syntax gate: target untouched
+OK    suite failure refused (exit 1)
+OK    suite failure: target reverted to clean
+OK    disabled policy refused (exit 2)
+OK    disabled policy: target untouched
+OK    failed run refused (exit 1)
+OK    failed run: target untouched
+OK    missing run exits 4 (exit 4)
+OK    invalid policy exits 2 (exit 2)
+
+selftest: 17 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-config.log b/agents/rocko/work/queue-73/out/gate/test-config.log
new file mode 100644
index 000000000..76c05ef74
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-config.log
@@ -0,0 +1,26 @@
+OK    absent adapter defaults to pi
+OK    adapter mock validates (exit 0)
+OK    unsupported adapter exits 2 (exit 2)
+OK    env exports adapter
+OK    bootstrap creates default when absent (exit 0)
+OK    bootstrap wrote config file
+OK    bootstrap is idempotent on existing config (exit 0)
+OK    bootstrap did not rewrite existing config
+OK    validate missing config exits 3 (exit 3)
+OK    malformed JSON exits 2 (exit 2)
+OK    unsupported configVersion exits 2 (exit 2)
+OK    unknown top-level key exits 2 (exit 2)
+OK    unknown execution key exits 2 (exit 2)
+OK    unsupported backend exits 2 (exit 2)
+OK    unsupported environment exits 2 (exit 2)
+OK    relative dataRoot exits 2 (exit 2)
+OK    non-canonical dataRoot exits 2 (exit 2)
+OK    filesystem root dataRoot exits 2 (exit 2)
+OK    home directory dataRoot exits 2 (exit 2)
+OK    dataRoot containing config dir exits 2 (exit 2)
+OK    control character in provider exits 2 (exit 2)
+OK    symlinked config file exits 2 (exit 2)
+OK    env exports resolve correctly
+OK    failed validation modified nothing
+
+selftest: 24 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-discord.log b/agents/rocko/work/queue-73/out/gate/test-discord.log
new file mode 100644
index 000000000..b013f9ffa
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-discord.log
@@ -0,0 +1,70 @@
+toolchain: node v26.8.1
+
+OK    syntax: packages/discord/src/approvals.mjs
+OK    syntax: packages/discord/src/authorize.mjs
+OK    syntax: packages/discord/src/binding.mjs
+OK    syntax: packages/discord/src/cli.mjs
+OK    syntax: packages/discord/src/connector.mjs
+OK    syntax: packages/discord/src/context.mjs
+OK    syntax: packages/discord/src/engine-pi.mjs
+OK    syntax: packages/discord/src/errors.mjs
+OK    syntax: packages/discord/src/gateway.mjs
+OK    syntax: packages/discord/src/git.mjs
+OK    syntax: packages/discord/src/journal.mjs
+OK    syntax: packages/discord/src/notify.mjs
+OK    syntax: packages/discord/src/rest.mjs
+OK    syntax: packages/discord/src/setspark.mjs
+OK    syntax: packages/discord/src/tools.mjs
+OK    syntax: packages/discord/src/web.mjs
+OK    syntax: packages/discord/bin/git-credential.mjs
+OK    syntax: packages/discord/extension/tools.mjs
+OK    syntax: packages/discord/tests/approvals.test.mjs
+OK    syntax: packages/discord/tests/authorize.test.mjs
+OK    syntax: packages/discord/tests/binding.test.mjs
+OK    syntax: packages/discord/tests/connector.test.mjs
+OK    syntax: packages/discord/tests/context.test.mjs
+OK    syntax: packages/discord/tests/engine.test.mjs
+OK    syntax: packages/discord/tests/fake-pi.mjs
+OK    syntax: packages/discord/tests/gateway.test.mjs
+OK    syntax: packages/discord/tests/git.test.mjs
+OK    syntax: packages/discord/tests/helpers.mjs
+OK    syntax: packages/discord/tests/journal.test.mjs
+OK    syntax: packages/discord/tests/notify.test.mjs
+OK    syntax: packages/discord/tests/recover.test.mjs
+OK    syntax: packages/discord/tests/rest.test.mjs
+OK    syntax: packages/discord/tests/setspark.test.mjs
+OK    syntax: packages/discord/tests/tools.test.mjs
+OK    syntax: packages/discord/tests/web.test.mjs
+OK    syntax: packages/discord/fixtures/claim-worker.mjs
+OK    syntax: packages/discord/fixtures/legacy-owner-worker.mjs
+OK    syntax: scripts/discord.sh
+OK    syntax: scripts/discord-service.sh
+OK    packages/discord declares no dependencies
+OK    no bot-token-shaped string in packages/discord
+OK    fixture binding uses placeholder ids only
+OK    fixture binding validates
+OK    real pi with the extension exposes exactly list_dir, read_file, search and no built-in tool
+OK    real pi with a writable root exposes exactly the three reads plus write_file and edit_file, and writes nothing at start
+OK    real pi with a web key exposes the three reads plus web_fetch and web_search, and no write tool without a writable root
+OK    real pi with a git root exposes the reads, writes and the four git verbs, commits nothing at start, and never shows the token
+OK    real pi with protocol vault adds reserve_id to the git verbs
+OK    real pi with a setspark key exposes the reads and the eight record verbs, no counters, and never shows the key
+OK    real pi refuses a git key on a read-only root (fail closed)
+OK    real pi with the pilot flags (--no-tools) exposes no tool at all
+OK    real pi exits non-zero without MOSAIC_DISCORD_TOOLS: no session, no tools (fail closed)
+OK    a failing nested test fails the run under a parent runner's NODE_TEST_CONTEXT
+OK    node --test packages/discord/tests/ (ℹ pass 178)
+OK    scripts/discord.sh --help exits 0
+OK    scripts/discord.sh check without a binding exits 4
+OK    scripts/discord.sh recover without a binding exits 4
+OK    scripts/discord.sh reload without a binding exits 4
+OK    scripts/discord-service.sh without a command exits 4
+OK    service unit renders with the repository path, a supervised run as the main process, exit 3 never retried, and reload as SIGHUP
+OK    service install writes the rendered unit (0644) and leaves no temp file
+OK    service install a second time reports unchanged
+OK    systemd-analyze verify accepts the rendered unit
+OK    service uninstall removes the unit file
+OK    service install with an unknown flag exits 4
+OK    service install with USER unset finishes and names the account for lingering
+
+discord suite: 66 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-extension-package.log b/agents/rocko/work/queue-73/out/gate/test-extension-package.log
new file mode 100644
index 000000000..40c37b8af
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-extension-package.log
@@ -0,0 +1,21 @@
+OK    initial ordinary-file install
+OK    installed tree matches canonical source
+OK    installed tree has no symlinks
+OK    check detects installation drift
+OK    sync refuses to overwrite installation drift
+OK    check detects an extra destination file
+OK    check detects an extra destination directory
+OK    check rejects a destination symlink
+OK    sync accepts a canonical source update
+OK    updated installation matches canonical source
+scripts/test-extension-package.sh: line 14: 231435 Killed                     "$@" > /dev/null 2>&1
+OK    forced interruption kills the replacing process
+OK    next invocation recovers old consistent installation
+OK    interrupted replacement rolled back
+OK    sync succeeds after interruption recovery
+OK    unlocked stale lock file does not block
+OK    active lock refuses a concurrent sync
+OK    source symlink fails closed
+OK    nested second entrypoint fails closed
+
+extension package selftest: 18 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-foundation.log b/agents/rocko/work/queue-73/out/gate/test-foundation.log
new file mode 100644
index 000000000..6338e1e6d
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-foundation.log
@@ -0,0 +1,53 @@
+toolchain: node v26.8.1, python 3.12.8, jsonschema 4.26.0
+
+OK    syntax: scripts/foundation-inspect.mjs
+OK    syntax: scripts/foundation/strict-json.mjs
+OK    syntax: scripts/foundation/canonical.mjs
+OK    syntax: scripts/foundation/resolve.mjs
+OK    syntax: scripts/foundation/validate-record.mjs
+OK    syntax: scripts/foundation/fixtures/build-fixtures.mjs
+OK    syntax: scripts/foundation/canonical.test.mjs
+OK    syntax: scripts/foundation/cli.test.mjs
+OK    syntax: scripts/foundation/fixtures.test.mjs
+OK    syntax: scripts/foundation/resolve.test.mjs
+OK    syntax: scripts/foundation/strict-json.test.mjs
+OK    syntax: scripts/foundation/verify-schema.py (ast only; no bytecode written)
+OK    fixture generator runs
+OK    checked-in fixtures/bundles equal a fresh generation
+OK    checked-in fixtures/raw equal a fresh generation
+OK    checked-in fixtures/index.json equal a fresh generation
+OK    checked-in demo bundles equal a fresh generation
+OK    a failing nested test fails the run under a parent runner's NODE_TEST_CONTEXT
+OK    node --test scripts/foundation/ (ℹ pass 80)
+OK    differential schema oracle: PASS: differential schema oracle (finite corpus; compatibility evidence, not equivalence proof)
+      platform witness: strftime('%Y') for year 999 -> '999' (pinned checker refuses years 0001..0999)
+      node v26.8.1; corpus 1568 records (38 pinned fixtures, 478 unique bundle records, 1052 typeCase/mutation/lexical cases)
+      schema column: agree-valid 540, agree-invalid 991, DISAGREEMENTS 0; strict-only (parser-bound) cases: 27; unsupported-kind records not schema-assessed by the inspector: 10
+      profile column (schema-valid records only): profile-valid 510, profile-invalid 30
+      profile refusals asserted: 30 schema-agreed-valid records refused only by the strict typed-string profile (rule profile-pattern-mismatch), 12 declared by name; 73 named probes verified against declared schema/profile columns
+OK    oracle: zero schema-column disagreements with the pinned checker
+OK    oracle: strict-only profile refusals are counted and asserted
+OK    demo: permitted read preview exits 0 (exit 0)
+OK    demo: permitted file.change preview exits 0 (exit 0)
+OK    demo: assignment.change proposal is unresolved (exit 3) (exit 3)
+OK    demo: revoked registration is refused (exit 3) (exit 3)
+OK    demo: message is not authority (exit 3) (exit 3)
+OK    usage: no arguments exits 2 (exit 2)
+OK    io: missing file exits 4 (exit 4)
+OK    io: directory exits 4 (exit 4)
+OK    io: symlink exits 4 (O_NOFOLLOW) (exit 4)
+OK    bound: oversize fixture exits 2 (exit 2)
+OK    profile: one final LF in a typed selection id is refused before admission (exit 2) (exit 2)
+OK    profile: two final LFs fail the schema pattern itself (exit 2) (exit 2)
+OK    profile: escaped newlines in free-form text stay allowed (exit 0) (exit 0)
+OK    profile refusal is invalid-request/profile-pattern-mismatch with selection and operation withheld, value not echoed
+OK    text output starts with the disclaimer
+OK    json output is valid JSON with result allowed and exactly the charter §7 fields
+OK    json golden matches byte-for-byte
+OK    sandboxed bundle run (env -i, PATH=/nonexistent) produced the unresolved proposal
+OK    sandbox inventory (path/type/size/mode/uid/gid/inode/mtime/sha256) unchanged by runs
+OK    canary never printed (bundle run and credential-file run)
+OK    a non-bundle JSON file is refused at the shape gate, not read into output
+OK    no field of the non-bundle file is echoed
+
+selftest: 44 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-queue.log b/agents/rocko/work/queue-73/out/gate/test-queue.log
new file mode 100644
index 000000000..3f0fc67d5
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-queue.log
@@ -0,0 +1,35 @@
+toolchain: node v26.8.1, git version 2.55.0
+
+OK    syntax: packages/queue/src/cli.mjs
+OK    syntax: packages/queue/src/errors.mjs
+OK    syntax: packages/queue/src/io.mjs
+OK    syntax: packages/queue/src/lock.mjs
+OK    syntax: packages/queue/src/queue.mjs
+OK    syntax: packages/queue/src/review.mjs
+OK    syntax: packages/queue/src/store.mjs
+OK    syntax: packages/queue/tests/commit.test.mjs
+OK    syntax: packages/queue/tests/data.test.mjs
+OK    syntax: packages/queue/tests/dispatch.test.mjs
+OK    syntax: packages/queue/tests/helpers.mjs
+OK    syntax: packages/queue/tests/lock.test.mjs
+OK    syntax: packages/queue/tests/migration.test.mjs
+OK    syntax: packages/queue/tests/review.test.mjs
+OK    syntax: packages/queue/tests/store.test.mjs
+OK    syntax: packages/queue/tests/write.test.mjs
+OK    syntax: packages/queue/tests/fixtures/fake-gitea.mjs
+OK    syntax: packages/queue/tests/fixtures/kill-at.mjs
+OK    syntax: packages/queue/tests/fixtures/lock-child.mjs
+OK    syntax: packages/queue/tests/fixtures/mosaic-pre-a2.sh
+OK    syntax: scripts/queue-commit.sh
+OK    syntax: scripts/git-hooks/pre-commit
+OK    syntax: scripts/mosaic
+OK    queue-commit.sh, the guard and scripts/mosaic are executable
+OK    packages/queue declares no dependencies
+ℹ tests 150
+ℹ pass 150
+ℹ fail 0
+OK    node --test packages/queue/tests/
+OK    scripts/mosaic queue help
+skip  queue verify and render --check: this checkout (/mnt/storage/scratch/rocko-r73) is not the queue's canonical root (/mnt/storage/src/mosaic-stack)
+
+queue suite: 27 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-release.log b/agents/rocko/work/queue-73/out/gate/test-release.log
new file mode 100644
index 000000000..7202b85d4
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-release.log
@@ -0,0 +1,16 @@
+OK    valid RELEASE resolves (exit 0)
+OK    invalid RELEASE exits 1 (exit 1)
+OK    missing RELEASE exits 1 (exit 1)
+OK    valid RELEASE leaves image tag consistent with version
+OK    status safe on empty state (exit 0)
+OK    status created no pointer
+OK    fault-injected activation refuses (exit 1)
+OK    refused activation wrote no pointer
+OK    refusal logged exactly once with valid fields
+OK    healthy activation succeeds (exit 0)
+OK    pointer written with valid fields
+OK    repeat activation succeeds (log grows) (exit 0)
+OK    log is append-only across activations
+OK    rollback without previous refuses (exit 1)
+
+selftest: 14 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/test-task.log b/agents/rocko/work/queue-73/out/gate/test-task.log
new file mode 100644
index 000000000..df6724a77
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/test-task.log
@@ -0,0 +1,101 @@
+OK    valid task validates (exit 0)
+OK    unknown task key exits 2 (exit 2)
+OK    unsupported taskVersion exits 2 (exit 2)
+OK    invalid task id exits 2 (exit 2)
+OK    empty prompt exits 2 (exit 2)
+OK    NUL in expectExact exits 2 (exit 2)
+OK    out-of-range timeout exits 2 (exit 2)
+OK    missing mission file exits 4 (exit 4)
+OK    task with valid mission validates (exit 0)
+OK    invalid mission exits 2 (exit 2)
+OK    validate missing task exits 4 (exit 4)
+OK    validation does not modify the task file
+OK    prune dry-run exits 0 (exit 0)
+OK    dry-run deleted nothing
+OK    prune --keep=2 --yes removes oldest (exit 0)
+OK    kept exactly 2 newest runs
+OK    newest run kept, oldest pruned
+OK    append-only receipt written (3 entries)
+OK    sessions/workspaces untouched by prune
+OK    prune with invalid keep exits 4 (exit 4)
+OK    mock adapter: gate passes on matching mock response (exit 0)
+OK    mock adapter: expect-mismatch recorded (exit 1)
+OK    mismatch reason recorded
+OK    unknown adapter fails closed (exit 1)
+OK    mission task runs via mock adapter (exit 0)
+OK    mission section injected into generated prompt
+OK    retry of mission run succeeds (exit 0)
+OK    retriedFrom lineage recorded
+OK    mission section present after retry (relative path resolved)
+OK    retry of missing run exits 4 (exit 4)
+OK    skill install bundled ms-tools (exit 0)
+OK    installed to skills-available
+OK    double install refuses (exit 1)
+OK    activate enables skill (exit 0)
+OK    enabled dir populated
+OK    uninstall while enabled refuses (exit 1)
+OK    deactivate moves back to available (exit 0)
+OK    uninstall removes from available (exit 0)
+OK    seat with enabled skill launches
+OK    skill path delivered to adapter
+OK    shipped researcher contract resolves (ceiling + network)
+OK    resolve-role refuses a non-role document (conductor policy) (exit 2)
+OK    resolve-role refuses name/filename mismatch (exit 2)
+OK    resolve-role refuses unknown network declaration (exit 2)
+OK    resolve-role refuses duplicate role tool (exit 2)
+OK    resolve-role refuses unsupported tool (exit 2)
+OK    resolve-role refuses missing contract file (exit 4)
+OK    version 1 role prints no contract line
+OK    shipped version 2 roles resolve with their contracts (pm, cto, coder, reviewer)
+OK    resolve-role accepts a minimal version 2 role (exit 0)
+OK    resolve-role refuses an action outside the vocabulary (exit 2)
+OK    resolve-role refuses a gated-only action in a role file (exit 2)
+OK    resolve-role refuses a Vikunja verb no role may hold (exit 2)
+OK    resolve-role refuses a version 2 role whose contract is missing (exit 2)
+OK    resolve-role refuses a contract that is a symbolic link (exit 2)
+OK    seat launches under role ceiling
+OK    role ceiling narrows seat tools (read,write,bash -> read,bash)
+OK    narrowing recorded loudly
+OK    missing role contract refuses launch (exit 2)
+OK    missing-contract refusal names the role
+OK    empty ceiling intersection -> tool-free seat
+OK    tool-free outcome recorded loudly
+OK    overridden agents dir without seat definition refuses (exit 4)
+OK    seat-resolution refusal names the seat
+OK    policy: mission only -> mission tools
+OK    policy: task only -> task tools
+OK    policy: both -> intersection (task narrowed)
+OK    policy: empty intersection -> tool-free
+OK    invalid mission capabilities rejected (exit 2)
+mosaic-task: unsupported mission tool: "sudo" (supported: read, write, edit, bash, grep, find, ls)
+OK    task run with user layer present (exit 0)
+OK    user context dispatched into generated prompt
+OK    workspace+tools task runs via mock (exit 0)
+OK    workspace path + tools delivered to adapter
+OK    persistent workspace created on host
+OK    plain task still runs (no workspace/tools) (exit 0)
+OK    workspace var present but empty when absent
+OK    tools empty when absent
+OK    unknown tool exits 2 (exit 2)
+OK    workspace traversal exits 2 (exit 2)
+OK    live hello task succeeds with exact marker
+OK    result.json written in run dir
+OK    result.json contents are correct
+OK    wrong expectExact fails with exit 1 (reason: expect-mismatch)
+OK    each run gets a distinct run dir (no clobber)
+OK    list shows both runs
+OK    fork base: teach succeeds (exit 0)
+OK    fork child recalls ancestor context (exit 0)
+OK    forked child recalled ancestor code word
+OK    ancestor session untouched by fork
+OK    child session dir has its own branch file
+OK    onboard without name exits 4 (non-interactive) (exit 4)
+OK    onboard --name renders profile (exit 0)
+OK    profile written
+OK    canon structure: required filled, optional placeholdered
+OK    canon sections present
+OK    user recall run succeeds
+OK    recalled user name (response: Jason)
+OK    no agent identity on headless run
+
+selftest: 98 passed, 0 failed
diff --git a/agents/rocko/work/queue-73/out/gate/webui.log b/agents/rocko/work/queue-73/out/gate/webui.log
new file mode 100644
index 000000000..ea5bcdef1
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/gate/webui.log
@@ -0,0 +1,59 @@
+✔ browser edge states: loading, empty, malformed, stale, hostile/long values, in-flight reply and appearance fallback (8106.572334ms)
+Rendered contrast: {"failures":[],"count":330,"lowest":4.504658476260286}
+✔ served Console browser: real board fixtures, keyboard, drafts, receipts, themes, 320px and failures (8747.634305ms)
+✔ S5 pages in a browser: real broker rows, inert text, Copy, routes, failures, layout and no writes (9887.359002ms)
+✔ bus routes serve the four reads from the reader view, unescaped JSON for the page to escape (264.222307ms)
+✔ decisions route serves the roles and recent lists from the reader view (109.024685ms)
+✔ bus routes are GET-only, have no write verb, and keep the same-origin checks (108.781146ms)
+✔ bus refusals map to statuses and never carry a path or a stack (82.034912ms)
+✔ humanCall speaks the human transport protocol without blocking the server (2363.712435ms)
+✔ busReader takes --business, else the live bus host, else refuses (837.234229ms)
+✔ humanCall kills a transport that runs past its timeout (595.698952ms)
+✔ serve refuses a --business value that is not a business id (1239.153702ms)
+✔ conversation view: full history, collapsed tools, hidden thinking, inert hostile content, malformed and reconcile markers (7658.493085ms)
+✔ conversation view: a fork keeps the open branch, says so, and opens the new one on request (2334.361368ms)
+✔ conversation view: a newer session with no readable history keeps the marker (1475.86245ms)
+✔ conversation view: seats without history say so and offer no reply (1154.68976ms)
+✔ Discord row through real board/WebUI: independent brake/liveness, no Reply, literal content (6505.513225ms)
+✔ return flow through the conversation view: send, tool call, delayed result, peer message, exact long answers, relaunch (57598.455344ms)
+✔ redactor: config, dataRoot, any absolute path, Discord ids and the JSON parser quote (2.787175ms)
+✔ credential state from the date alone, as the bus states it (0.265543ms)
+✔ queue: rows, one row, ids and methods, notes; no path or id from a row (2447.752007ms)
+✔ queue: a refused read fails closed with the store's text, redacted (1106.728316ms)
+✔ queue: a store that fails to load, or a child that dies, gives a fixed message with no path or stack (2952.199165ms)
+✔ business: names, vars on the allowlist, authority, credential metadata only (43.48209ms)
+✔ business: missing or invalid file refuses with the module's text, redacted (18.270212ms)
+✔ business without --business: the live bus host, else no-bus-host; no system config: not-configured (23.291575ms)
+✔ settings: release, board, loopback address, notifier binding name only (657.992106ms)
+✔ both presentations replace old activity with relaunch notice, label retained history, then resume after new activity (7673.38997ms)
+✔ reported return flow and relative Age: reply sent from the inspector, then the new answer appears there without manual refresh (26157.798956ms)
+✔ loopback host and board origin fail closed (12.188559ms)
+✔ real board fixture passes through WebUI; assets and isolated seen/reply work (347.900023ms)
+✔ proxy preserves exact request bytes, status and receipt, rejects forms and malformed JSON, never follows redirect (276.644747ms)
+✔ unreachable board reports URL; CLI rejects unsupported options (1741.484661ms)
+✔ render prints the unit without the preview flags (60.324721ms)
+✔ render carries every serve flag, the preview set included (85.287597ms)
+✔ a partial preview set is refused (184.902743ms)
+✔ a value the unit cannot carry is refused, for every flag (3027.921779ms)
+✔ the rendered unit names none of the human proof agent markers and sets no agent variable (130.32085ms)
+✔ install into --dir with --no-reload writes the file whole, then reports unchanged (738.374793ms)
+✔ install without --no-reload runs daemon-reload and nothing that starts the unit (114.539901ms)
+✔ uninstall refuses while the unit is active, and removes it once it is not (669.457415ms)
+✔ status prints the unit state and GET /healthz on the installed port (700.129598ms)
+✔ usage: an unknown verb or flag, and start or enable, exit 4 (267.924672ms)
+✔ tokens.css and icons.svg are byte copies of docs/design (1.609561ms)
+✔ JetBrains Mono files match the digests in their sources file, with the OFL text (3.068117ms)
+✔ shipped strings are unchanged, and the web keeps its write boundary (2.504909ms)
+✔ shell in a browser: System default, seeded states, 400px, focus and the Ctrl+K palette (12906.255137ms)
+✔ five states on Board, Inbox, Tasks, Agents and Trail (2975.960387ms)
+✔ Queue, Business and Settings in their five states, with no move button (10184.597385ms)
+✔ seeded secrets, Discord ids and paths reach no response and no page; business refusals in the module's words (2343.523176ms)
+✔ the server log names no config path when the system config refuses (580.596691ms)
+ℹ tests 50
+ℹ suites 0
+ℹ pass 50
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 58347.310973
diff --git a/agents/rocko/work/queue-73/out/mutants-base.txt b/agents/rocko/work/queue-73/out/mutants-base.txt
new file mode 100644
index 000000000..264a588af
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base.txt
@@ -0,0 +1,6 @@
+baseline: pass
+role-keeps-reassign: NOT APPLIED (sed matched nothing)
+broker-unlisted-cross: SURVIVED
+resolve-cross-unnarrowed: killed (✖ an action outside the instance's authority has no tool;✖ limits.authority narrows cross-role actions too;)
+classify-unlisted-cross: killed (✖ classify;✖ limits.authority narrows cross-role actions too;✖ limits.authority without role.launch leaves the launcher with no launch block;✖ limits narrow the definition and never widen it;✖ role.launch stays within-role only for the instance the launch block names;)
+tools-all-task-verbs: killed (✖ an action outside the instance's authority has no tool;)
diff --git a/agents/rocko/work/queue-73/out/mutants-base/baseline.log b/agents/rocko/work/queue-73/out/mutants-base/baseline.log
new file mode 100644
index 000000000..bb983fa4a
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/baseline.log
@@ -0,0 +1,67 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (41.10989ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (30.340056ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (27.301914ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (16.784646ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (16.117365ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (16.007796ms)
+✔ task action subjects and linked decision trail are complete and ordered (33.402123ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (12.499912ms)
+✔ business isolation includes inherited object names and cross-business message references (19.585889ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (27.095845ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (12.374754ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (21.007289ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (14.896337ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (21.130329ms)
+✔ systemVars flattens the validated config (3.298819ms)
+✔ precedence: system, business, project, project role, agent (9.974934ms)
+✔ limits narrow the definition and never widen it (4.037503ms)
+✔ role.launch stays within-role only for the instance the launch block names (10.06119ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (3.498072ms)
+✔ limits.authority narrows cross-role actions too (1.527146ms)
+✔ classify (1.816472ms)
+✔ the record carries what the broker and launcher need (1.558999ms)
+✔ digest: key order doesn't matter, any value change does (10.199487ms)
+✔ refusals (4.176465ms)
+✔ the four shipped version 2 roles load (14.372215ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.984217ms)
+✔ shipped authority follows the note's table (1.084755ms)
+✔ version 1 files keep loading with no authority (2.030792ms)
+✔ the conductor policy isn't a role (0.422745ms)
+✔ a missing role file is exit 4, a symbolic link too (0.604689ms)
+✔ version 2 refusals (2.11256ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (3.892762ms)
+✔ credentials: Gitea scopes (1.509172ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (2.083022ms)
+✔ credentials: services (0.971694ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.069304ms)
+✔ the PM gets launch, its task verbs and the reads (10.141611ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (3.092654ms)
+✔ launch only when the business's launch block names the instance as launcher (3.195221ms)
+✔ an action outside the instance's authority has no tool (2.493976ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (17.882552ms)
+✔ task.create needs a recorded human request and a requirement id (62.327198ms)
+✔ only labels named in the business file can be written (37.091411ms)
+✔ task.schedule sets and clears a due date and relations (44.451536ms)
+✔ assign and reassign move the role bots and record task.assigned (66.030754ms)
+✔ task.update.assigned is for the assignee and records task.state (66.783287ms)
+✔ a wrong expected digest records task.conflict and writes nothing (27.559383ms)
+✔ a cross-role verb needs a resolved decision, used once (41.3223ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (39.501253ms)
+✔ a lost answer is settled by a re-read and never retried (48.948392ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (45.537621ms)
+✔ a task the sync bot cannot read refuses and records nothing (24.722852ms)
+✔ verbs and polls for one business run one at a time (107.598043ms)
+✔ a due date with milliseconds is written to the second (99.477546ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (51.18486ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (30.29597ms)
+✔ a create whose final read fails succeeds and records task.created (43.712142ms)
+✔ an edit between the last write and the final read shows as external on the next poll (60.844534ms)
+✔ task.created is recorded when a later label write fails (43.90553ms)
+ℹ tests 59
+ℹ suites 0
+ℹ pass 59
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1118.992245
diff --git a/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.diff b/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.diff
new file mode 100644
index 000000000..b0bdc2685
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-base/orig	2026-10-11 18:09:25.811161905 -0500
++++ packages/bus/src/broker.mjs	2026-10-11 18:09:27.269340616 -0500
+@@ -274,7 +274,7 @@
+       ? 'within-role'
+       : a?.crossRole.includes(action)
+         ? 'cross-role'
+-        : 'gated';
++        : 'cross-role';
+   }
+   #decision(s, id) {
+     const d = this.#store.get('SELECT * FROM decisions WHERE id=? AND business=?', id, s.business);
diff --git a/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.log b/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.log
new file mode 100644
index 000000000..9a4393c1b
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/broker-unlisted-cross.log
@@ -0,0 +1,67 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (52.058368ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (31.827945ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (32.343445ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (15.453301ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (16.112223ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (15.465475ms)
+✔ task action subjects and linked decision trail are complete and ordered (34.817745ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (11.489008ms)
+✔ business isolation includes inherited object names and cross-business message references (25.296217ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (31.184317ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (16.792175ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (22.551334ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (13.915971ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (23.619893ms)
+✔ systemVars flattens the validated config (3.200801ms)
+✔ precedence: system, business, project, project role, agent (7.711976ms)
+✔ limits narrow the definition and never widen it (5.662494ms)
+✔ role.launch stays within-role only for the instance the launch block names (11.879594ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (4.345739ms)
+✔ limits.authority narrows cross-role actions too (1.909662ms)
+✔ classify (2.20486ms)
+✔ the record carries what the broker and launcher need (2.46906ms)
+✔ digest: key order doesn't matter, any value change does (16.682292ms)
+✔ refusals (3.683678ms)
+✔ the four shipped version 2 roles load (5.36865ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.795612ms)
+✔ shipped authority follows the note's table (1.155071ms)
+✔ version 1 files keep loading with no authority (1.552767ms)
+✔ the conductor policy isn't a role (0.342392ms)
+✔ a missing role file is exit 4, a symbolic link too (0.602196ms)
+✔ version 2 refusals (2.146247ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (3.688492ms)
+✔ credentials: Gitea scopes (2.236086ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (2.557229ms)
+✔ credentials: services (0.936797ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.47412ms)
+✔ the PM gets launch, its task verbs and the reads (11.865141ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (4.766334ms)
+✔ launch only when the business's launch block names the instance as launcher (3.204748ms)
+✔ an action outside the instance's authority has no tool (2.992598ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (16.830589ms)
+✔ task.create needs a recorded human request and a requirement id (67.154682ms)
+✔ only labels named in the business file can be written (33.442526ms)
+✔ task.schedule sets and clears a due date and relations (48.841556ms)
+✔ assign and reassign move the role bots and record task.assigned (78.987774ms)
+✔ task.update.assigned is for the assignee and records task.state (65.470032ms)
+✔ a wrong expected digest records task.conflict and writes nothing (32.691954ms)
+✔ a cross-role verb needs a resolved decision, used once (37.492507ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (45.183086ms)
+✔ a lost answer is settled by a re-read and never retried (66.58075ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (47.459336ms)
+✔ a task the sync bot cannot read refuses and records nothing (24.247305ms)
+✔ verbs and polls for one business run one at a time (102.11683ms)
+✔ a due date with milliseconds is written to the second (77.242581ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (48.960103ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (20.384311ms)
+✔ a create whose final read fails succeeds and records task.created (40.019605ms)
+✔ an edit between the last write and the final read shows as external on the next poll (48.872647ms)
+✔ task.created is recorded when a later label write fails (28.904685ms)
+ℹ tests 59
+ℹ suites 0
+ℹ pass 59
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1099.85091
diff --git a/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.diff b/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.diff
new file mode 100644
index 000000000..6ace5fb95
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.diff
@@ -0,0 +1,9 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-base/orig	2026-10-11 18:09:25.812634433 -0500
++++ packages/business/src/resolve.mjs	2026-10-11 18:09:29.450323415 -0500
+@@ -100,5 +100,5 @@
+   if (!ACTIONS.includes(action)) refuse(`unknown action: ${JSON.stringify(action)}`);
+   if (resolved.limits.authority.withinRole.includes(action)) return "within";
+   if (resolved.limits.authority.crossRole.includes(action)) return "cross";
+-  return "gated";
++  return "cross";
+ }
diff --git a/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.log b/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.log
new file mode 100644
index 000000000..65310504e
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/classify-unlisted-cross.log
@@ -0,0 +1,174 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (34.562762ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (48.881252ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (31.40005ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (19.163154ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (18.637808ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (23.106973ms)
+✔ task action subjects and linked decision trail are complete and ordered (33.924385ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (12.548753ms)
+✔ business isolation includes inherited object names and cross-business message references (18.991923ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (21.671041ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (12.2862ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (22.485066ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (14.168727ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (23.026ms)
+✔ systemVars flattens the validated config (3.055626ms)
+✔ precedence: system, business, project, project role, agent (7.376521ms)
+✖ limits narrow the definition and never widen it (4.765621ms)
+✖ role.launch stays within-role only for the instance the launch block names (5.845037ms)
+✖ limits.authority without role.launch leaves the launcher with no launch block (2.382717ms)
+✖ limits.authority narrows cross-role actions too (2.320376ms)
+✖ classify (1.91099ms)
+✔ the record carries what the broker and launcher need (1.859849ms)
+✔ digest: key order doesn't matter, any value change does (16.863167ms)
+✔ refusals (4.843702ms)
+✔ the four shipped version 2 roles load (4.952568ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.790725ms)
+✔ shipped authority follows the note's table (0.859956ms)
+✔ version 1 files keep loading with no authority (1.580548ms)
+✔ the conductor policy isn't a role (0.410542ms)
+✔ a missing role file is exit 4, a symbolic link too (0.687476ms)
+✔ version 2 refusals (2.208359ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (4.089476ms)
+✔ credentials: Gitea scopes (1.522708ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.96425ms)
+✔ credentials: services (1.096518ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.388238ms)
+✔ the PM gets launch, its task verbs and the reads (10.881554ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (4.91844ms)
+✔ launch only when the business's launch block names the instance as launcher (2.979843ms)
+✔ an action outside the instance's authority has no tool (2.821761ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (16.298799ms)
+✔ task.create needs a recorded human request and a requirement id (77.423799ms)
+✔ only labels named in the business file can be written (39.818018ms)
+✔ task.schedule sets and clears a due date and relations (52.774716ms)
+✔ assign and reassign move the role bots and record task.assigned (70.210371ms)
+✔ task.update.assigned is for the assignee and records task.state (71.320218ms)
+✔ a wrong expected digest records task.conflict and writes nothing (24.030222ms)
+✔ a cross-role verb needs a resolved decision, used once (35.582912ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (33.762759ms)
+✔ a lost answer is settled by a re-read and never retried (33.963244ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (49.791704ms)
+✔ a task the sync bot cannot read refuses and records nothing (28.959016ms)
+✔ verbs and polls for one business run one at a time (103.849164ms)
+✔ a due date with milliseconds is written to the second (77.623434ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (41.880183ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (18.077862ms)
+✔ a create whose final read fails succeeds and records task.created (38.538489ms)
+✔ an edit between the last write and the final read shows as external on the next poll (50.303392ms)
+✔ task.created is recorded when a later label write fails (21.206701ms)
+ℹ tests 59
+ℹ suites 0
+ℹ pass 54
+ℹ fail 5
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1016.65102
+
+✖ failing tests:
+
+test at packages/business/tests/resolve.test.mjs:69:1
+✖ limits narrow the definition and never widen it (4.765621ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:87:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:91:1
+✖ role.launch stays within-role only for the instance the launch block names (5.845037ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:98:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:112:1
+✖ limits.authority without role.launch leaves the launcher with no launch block (2.382717ms)
+  AssertionError [ERR_ASSERTION]: agent
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:119:12)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: false,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:124:1
+✖ limits.authority narrows cross-role actions too (2.320376ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:130:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:135:1
+✖ classify (1.91099ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:140:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.diff b/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.diff
new file mode 100644
index 000000000..77d60bb53
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-base/orig	2026-10-11 18:09:25.812634433 -0500
++++ packages/business/src/resolve.mjs	2026-10-11 18:09:28.444840244 -0500
+@@ -65,7 +65,7 @@
+   if (vars["limits.network"]) network = narrowerNetwork(network, vars["limits.network"]);
+   if (vars["limits.authority"]) {
+     withinRole = withinRole.filter((a) => vars["limits.authority"].includes(a));
+-    crossRole = crossRole.filter((a) => vars["limits.authority"].includes(a));
++    crossRole = [...crossRole];
+   }
+   // role.launch needs the business file's launch block naming this
+   // instance (addendum A section 8), and the verb must survive
diff --git a/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.log b/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.log
new file mode 100644
index 000000000..c32d8d957
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/resolve-cross-unnarrowed.log
@@ -0,0 +1,127 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (32.184603ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (34.006489ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (38.37318ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (17.069074ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (18.894895ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (14.699724ms)
+✔ task action subjects and linked decision trail are complete and ordered (34.571108ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (11.623471ms)
+✔ business isolation includes inherited object names and cross-business message references (18.142067ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (23.301159ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (9.613466ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (19.23009ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (12.482528ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (19.85656ms)
+✔ systemVars flattens the validated config (3.10758ms)
+✔ precedence: system, business, project, project role, agent (8.111606ms)
+✔ limits narrow the definition and never widen it (4.340763ms)
+✔ role.launch stays within-role only for the instance the launch block names (7.7091ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (4.788773ms)
+✖ limits.authority narrows cross-role actions too (4.913175ms)
+✔ classify (2.31828ms)
+✔ the record carries what the broker and launcher need (1.866404ms)
+✔ digest: key order doesn't matter, any value change does (11.152959ms)
+✔ refusals (4.229422ms)
+✔ the four shipped version 2 roles load (4.850962ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.823772ms)
+✔ shipped authority follows the note's table (0.832213ms)
+✔ version 1 files keep loading with no authority (1.502179ms)
+✔ the conductor policy isn't a role (0.338784ms)
+✔ a missing role file is exit 4, a symbolic link too (0.574289ms)
+✔ version 2 refusals (1.813974ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (3.519361ms)
+✔ credentials: Gitea scopes (1.479096ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.986778ms)
+✔ credentials: services (0.824537ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.067344ms)
+✔ the PM gets launch, its task verbs and the reads (12.570361ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (4.699967ms)
+✔ launch only when the business's launch block names the instance as launcher (3.180079ms)
+✖ an action outside the instance's authority has no tool (6.26463ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (16.396169ms)
+✔ task.create needs a recorded human request and a requirement id (68.567729ms)
+✔ only labels named in the business file can be written (37.772712ms)
+✔ task.schedule sets and clears a due date and relations (42.822471ms)
+✔ assign and reassign move the role bots and record task.assigned (65.432289ms)
+✔ task.update.assigned is for the assignee and records task.state (62.852511ms)
+✔ a wrong expected digest records task.conflict and writes nothing (28.459203ms)
+✔ a cross-role verb needs a resolved decision, used once (28.527583ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (28.982719ms)
+✔ a lost answer is settled by a re-read and never retried (30.392592ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (38.949964ms)
+✔ a task the sync bot cannot read refuses and records nothing (25.968994ms)
+✔ verbs and polls for one business run one at a time (99.920681ms)
+✔ a due date with milliseconds is written to the second (69.877437ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (40.473302ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (16.49068ms)
+✔ a create whose final read fails succeeds and records task.created (38.315534ms)
+✔ an edit between the last write and the final read shows as external on the next poll (45.052755ms)
+✔ task.created is recorded when a later label write fails (19.326478ms)
+ℹ tests 59
+ℹ suites 0
+ℹ pass 57
+ℹ fail 2
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 943.487341
+
+✖ failing tests:
+
+test at packages/business/tests/resolve.test.mjs:124:1
+✖ limits.authority narrows cross-role actions too (4.913175ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    {
+      crossRole: [
+  +     'task.reassign',
+        'task.scope.change'
+      ],
+      withinRole: [
+        'task.update.assigned',
+        'git.push.working'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/business/tests/resolve.test.mjs:129:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: { withinRole: [ 'task.update.assigned', 'git.push.working' ], crossRole: [ 'task.reassign', 'task.scope.change' ] },
+    expected: { withinRole: [ 'task.update.assigned', 'git.push.working' ], crossRole: [ 'task.scope.change' ] },
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
+
+test at packages/harness/tests/tools.test.mjs:44:1
+✖ an action outside the instance's authority has no tool (6.26463ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+  +   'task_reassign',
+  +   'task_scope_change',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+      'trail',
+      'list_tasks'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/harness/tests/tools.test.mjs:46:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'task_reassign', 'task_scope_change', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.diff b/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.diff
new file mode 100644
index 000000000..a67cdc1f4
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-base/orig	2026-10-11 18:09:25.821593099 -0500
++++ packages/harness/src/tools.mjs	2026-10-11 18:09:30.532803215 -0500
+@@ -89,7 +89,7 @@
+   }
+   if (tracker) {
+     for (const [action, [name, description, properties, required]] of Object.entries(TASK_TOOLS)) {
+-      if (held.has(action)) add(name, action, action, description, obj(properties, required));
++      add(name, action, action, description, obj(properties, required));
+     }
+   }
+   add("raise_decision", "decision.raise", null, "Ask for a decision before a gated or cross-role action, or when you need a choice made. The answer arrives as a message.", obj({
diff --git a/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.log b/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.log
new file mode 100644
index 000000000..63e038c14
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-base/tools-all-task-verbs.log
@@ -0,0 +1,104 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (68.304242ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (36.220226ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (28.451192ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (17.620543ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (17.134783ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (17.523052ms)
+✔ task action subjects and linked decision trail are complete and ordered (28.196094ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (11.654167ms)
+✔ business isolation includes inherited object names and cross-business message references (17.773328ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (63.640909ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (30.76888ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (72.884982ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (38.656622ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (86.595811ms)
+✔ systemVars flattens the validated config (3.418009ms)
+✔ precedence: system, business, project, project role, agent (15.12961ms)
+✔ limits narrow the definition and never widen it (32.983425ms)
+✔ role.launch stays within-role only for the instance the launch block names (11.91897ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (2.882384ms)
+✔ limits.authority narrows cross-role actions too (3.171349ms)
+✔ classify (3.090892ms)
+✔ the record carries what the broker and launcher need (2.902871ms)
+✔ digest: key order doesn't matter, any value change does (14.329947ms)
+✔ refusals (3.553461ms)
+✔ the four shipped version 2 roles load (6.936567ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.580417ms)
+✔ shipped authority follows the note's table (10.829435ms)
+✔ version 1 files keep loading with no authority (1.892986ms)
+✔ the conductor policy isn't a role (0.383968ms)
+✔ a missing role file is exit 4, a symbolic link too (0.600117ms)
+✔ version 2 refusals (1.966349ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (4.284811ms)
+✔ credentials: Gitea scopes (1.537433ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.912443ms)
+✔ credentials: services (1.044154ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.05934ms)
+✔ the PM gets launch, its task verbs and the reads (12.754906ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (4.823918ms)
+✔ launch only when the business's launch block names the instance as launcher (3.337282ms)
+✖ an action outside the instance's authority has no tool (6.565686ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (19.618993ms)
+✔ task.create needs a recorded human request and a requirement id (57.310658ms)
+✔ only labels named in the business file can be written (37.756552ms)
+✔ task.schedule sets and clears a due date and relations (44.948474ms)
+✔ assign and reassign move the role bots and record task.assigned (133.26048ms)
+✔ task.update.assigned is for the assignee and records task.state (149.616551ms)
+✔ a wrong expected digest records task.conflict and writes nothing (60.89642ms)
+✔ a cross-role verb needs a resolved decision, used once (61.058916ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (43.717227ms)
+✔ a lost answer is settled by a re-read and never retried (60.615688ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (50.786839ms)
+✔ a task the sync bot cannot read refuses and records nothing (33.128318ms)
+✔ verbs and polls for one business run one at a time (106.542625ms)
+✔ a due date with milliseconds is written to the second (89.23896ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (52.735737ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (38.534275ms)
+✔ a create whose final read fails succeeds and records task.created (48.772964ms)
+✔ an edit between the last write and the final read shows as external on the next poll (64.248179ms)
+✔ task.created is recorded when a later label write fails (23.685508ms)
+ℹ tests 59
+ℹ suites 0
+ℹ pass 58
+ℹ fail 1
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 1367.254328
+
+✖ failing tests:
+
+test at packages/harness/tests/tools.test.mjs:44:1
+✖ an action outside the instance's authority has no tool (6.565686ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+  +   'task_create',
+  +   'task_assign',
+  +   'task_reassign',
+  +   'task_schedule',
+  +   'task_priority_change',
+  +   'task_scope_change',
+  +   'task_update_assigned',
+  +   'task_close',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+      'trail',
+      'list_tasks'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73-base/packages/harness/tests/tools.test.mjs:46:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'task_create', 'task_assign', 'task_reassign', 'task_schedule', 'task_priority_change', 'task_scope_change', 'task_update_assigned', 'task_close', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate.txt b/agents/rocko/work/queue-73/out/mutants-candidate.txt
new file mode 100644
index 000000000..6efae2c74
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate.txt
@@ -0,0 +1,6 @@
+baseline: pass
+role-keeps-reassign: killed (✖ a coder with a tracker gets its own task tools and no task_reassign;✖ classify;✖ shipped authority follows the note's table;✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role;)
+broker-unlisted-cross: killed (✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role;)
+resolve-cross-unnarrowed: killed (✖ an action outside the instance's authority has no tool;✖ limits.authority narrows cross-role actions too;)
+classify-unlisted-cross: killed (✖ classify;✖ limits.authority narrows cross-role actions too;✖ limits.authority without role.launch leaves the launcher with no launch block;✖ limits narrow the definition and never widen it;✖ role.launch stays within-role only for the instance the launch block names;)
+tools-all-task-verbs: killed (✖ a coder with a tracker gets its own task tools and no task_reassign;✖ an action outside the instance's authority has no tool;)
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/baseline.log b/agents/rocko/work/queue-73/out/mutants-candidate/baseline.log
new file mode 100644
index 000000000..5ecd3b74a
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/baseline.log
@@ -0,0 +1,69 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (22.59035ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (26.760387ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (25.941999ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (14.175334ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (18.000242ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (17.704479ms)
+✔ task action subjects and linked decision trail are complete and ordered (27.462393ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (11.228427ms)
+✔ business isolation includes inherited object names and cross-business message references (18.813263ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (22.467431ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (10.563551ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (15.243023ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (13.037666ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (24.729793ms)
+✔ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (19.098206ms)
+✔ systemVars flattens the validated config (1.823722ms)
+✔ precedence: system, business, project, project role, agent (7.082391ms)
+✔ limits narrow the definition and never widen it (4.101089ms)
+✔ role.launch stays within-role only for the instance the launch block names (7.044571ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (3.176026ms)
+✔ limits.authority narrows cross-role actions too (3.958115ms)
+✔ classify (1.566137ms)
+✔ the record carries what the broker and launcher need (1.679601ms)
+✔ digest: key order doesn't matter, any value change does (9.40251ms)
+✔ refusals (3.314557ms)
+✔ the four shipped version 2 roles load (4.562073ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.373138ms)
+✔ shipped authority follows the note's table (0.880774ms)
+✔ version 1 files keep loading with no authority (1.340018ms)
+✔ the conductor policy isn't a role (0.294463ms)
+✔ a missing role file is exit 4, a symbolic link too (0.573454ms)
+✔ version 2 refusals (2.142424ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (2.806218ms)
+✔ credentials: Gitea scopes (1.390039ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.74487ms)
+✔ credentials: services (1.066435ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.503718ms)
+✔ the PM gets launch, its task verbs and the reads (7.458116ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.543347ms)
+✔ a coder with a tracker gets its own task tools and no task_reassign (2.204646ms)
+✔ launch only when the business's launch block names the instance as launcher (1.84137ms)
+✔ an action outside the instance's authority has no tool (2.443015ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (11.834983ms)
+✔ task.create needs a recorded human request and a requirement id (52.551332ms)
+✔ only labels named in the business file can be written (36.191609ms)
+✔ task.schedule sets and clears a due date and relations (42.137434ms)
+✔ assign and reassign move the role bots and record task.assigned (61.839072ms)
+✔ task.update.assigned is for the assignee and records task.state (64.945829ms)
+✔ a wrong expected digest records task.conflict and writes nothing (26.65454ms)
+✔ a cross-role verb needs a resolved decision, used once (33.944638ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (33.109479ms)
+✔ a lost answer is settled by a re-read and never retried (34.373385ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (39.628552ms)
+✔ a task the sync bot cannot read refuses and records nothing (18.56014ms)
+✔ verbs and polls for one business run one at a time (99.689243ms)
+✔ a due date with milliseconds is written to the second (72.921801ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (43.896182ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (18.194316ms)
+✔ a create whose final read fails succeeds and records task.created (39.258208ms)
+✔ an edit between the last write and the final read shows as external on the next poll (42.888728ms)
+✔ task.created is recorded when a later label write fails (17.415777ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 61
+ℹ fail 0
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 906.528938
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.diff b/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.diff
new file mode 100644
index 000000000..be576a401
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-candidate/orig	2026-10-11 18:07:23.086216318 -0500
++++ packages/bus/src/broker.mjs	2026-10-11 18:14:56.826892306 -0500
+@@ -274,7 +274,7 @@
+       ? 'within-role'
+       : a?.crossRole.includes(action)
+         ? 'cross-role'
+-        : 'gated';
++        : 'cross-role';
+   }
+   #decision(s, id) {
+     const d = this.#store.get('SELECT * FROM decisions WHERE id=? AND business=?', id, s.business);
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.log b/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.log
new file mode 100644
index 000000000..35622fbbf
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/broker-unlisted-cross.log
@@ -0,0 +1,94 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (28.422584ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (25.633236ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (21.236303ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (14.292941ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (16.245186ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (14.466042ms)
+✔ task action subjects and linked decision trail are complete and ordered (24.311959ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (9.415662ms)
+✔ business isolation includes inherited object names and cross-business message references (17.125506ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (20.297014ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (11.125627ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (17.296304ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (11.986569ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (19.617388ms)
+✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (18.46808ms)
+✔ systemVars flattens the validated config (3.618153ms)
+✔ precedence: system, business, project, project role, agent (7.398492ms)
+✔ limits narrow the definition and never widen it (3.541406ms)
+✔ role.launch stays within-role only for the instance the launch block names (7.555402ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (3.395477ms)
+✔ limits.authority narrows cross-role actions too (1.804949ms)
+✔ classify (2.023344ms)
+✔ the record carries what the broker and launcher need (1.416391ms)
+✔ digest: key order doesn't matter, any value change does (7.582607ms)
+✔ refusals (6.552895ms)
+✔ the four shipped version 2 roles load (3.303484ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.217421ms)
+✔ shipped authority follows the note's table (0.789951ms)
+✔ version 1 files keep loading with no authority (1.087748ms)
+✔ the conductor policy isn't a role (0.215847ms)
+✔ a missing role file is exit 4, a symbolic link too (0.407959ms)
+✔ version 2 refusals (1.339954ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (2.502025ms)
+✔ credentials: Gitea scopes (1.02232ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.91448ms)
+✔ credentials: services (0.677159ms)
+✔ contract: a non-empty regular Markdown file beside the role file (0.758295ms)
+✔ the PM gets launch, its task verbs and the reads (7.934889ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.113216ms)
+✔ a coder with a tracker gets its own task tools and no task_reassign (1.793071ms)
+✔ launch only when the business's launch block names the instance as launcher (1.614854ms)
+✔ an action outside the instance's authority has no tool (4.100632ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (10.303855ms)
+✔ task.create needs a recorded human request and a requirement id (43.05711ms)
+✔ only labels named in the business file can be written (27.611979ms)
+✔ task.schedule sets and clears a due date and relations (39.061519ms)
+✔ assign and reassign move the role bots and record task.assigned (62.22098ms)
+✔ task.update.assigned is for the assignee and records task.state (55.970065ms)
+✔ a wrong expected digest records task.conflict and writes nothing (25.10594ms)
+✔ a cross-role verb needs a resolved decision, used once (27.276288ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (27.716914ms)
+✔ a lost answer is settled by a re-read and never retried (29.052452ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (39.47631ms)
+✔ a task the sync bot cannot read refuses and records nothing (20.196553ms)
+✔ verbs and polls for one business run one at a time (92.77395ms)
+✔ a due date with milliseconds is written to the second (67.659021ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (39.62171ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (16.201161ms)
+✔ a create whose final read fails succeeds and records task.created (35.582005ms)
+✔ an edit between the last write and the final read shows as external on the next poll (44.800687ms)
+✔ task.created is recorded when a later label write fails (15.946503ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 60
+ℹ fail 1
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 829.410177
+
+✖ failing tests:
+
+test at packages/bus/tests/broker.test.mjs:361:1
+✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (18.46808ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  + actual - expected
+  
+  + 'cross-role'
+  - 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/bus/tests/broker.test.mjs:374:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross-role',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.diff b/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.diff
new file mode 100644
index 000000000..1ca1bd35d
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.diff
@@ -0,0 +1,9 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-candidate/orig	2026-10-11 18:07:23.087807382 -0500
++++ packages/business/src/resolve.mjs	2026-10-11 18:14:58.604372993 -0500
+@@ -100,5 +100,5 @@
+   if (!ACTIONS.includes(action)) refuse(`unknown action: ${JSON.stringify(action)}`);
+   if (resolved.limits.authority.withinRole.includes(action)) return "within";
+   if (resolved.limits.authority.crossRole.includes(action)) return "cross";
+-  return "gated";
++  return "cross";
+ }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.log b/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.log
new file mode 100644
index 000000000..7545a7a94
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/classify-unlisted-cross.log
@@ -0,0 +1,176 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (32.115093ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (20.806345ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (20.504225ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (15.031318ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (13.795536ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (12.743478ms)
+✔ task action subjects and linked decision trail are complete and ordered (20.909183ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (11.281359ms)
+✔ business isolation includes inherited object names and cross-business message references (14.550486ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (18.52182ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (11.841502ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (16.416647ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (10.122398ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (18.177921ms)
+✔ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (16.866523ms)
+✔ systemVars flattens the validated config (2.898361ms)
+✔ precedence: system, business, project, project role, agent (7.459935ms)
+✖ limits narrow the definition and never widen it (4.478519ms)
+✖ role.launch stays within-role only for the instance the launch block names (5.021168ms)
+✖ limits.authority without role.launch leaves the launcher with no launch block (1.627639ms)
+✖ limits.authority narrows cross-role actions too (1.276774ms)
+✖ classify (1.320238ms)
+✔ the record carries what the broker and launcher need (1.382509ms)
+✔ digest: key order doesn't matter, any value change does (7.774975ms)
+✔ refusals (2.214875ms)
+✔ the four shipped version 2 roles load (3.34149ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.017705ms)
+✔ shipped authority follows the note's table (0.534057ms)
+✔ version 1 files keep loading with no authority (0.917276ms)
+✔ the conductor policy isn't a role (0.211232ms)
+✔ a missing role file is exit 4, a symbolic link too (0.388821ms)
+✔ version 2 refusals (1.302921ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (2.052877ms)
+✔ credentials: Gitea scopes (0.840766ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.03215ms)
+✔ credentials: services (0.560366ms)
+✔ contract: a non-empty regular Markdown file beside the role file (0.659056ms)
+✔ the PM gets launch, its task verbs and the reads (7.878729ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.195081ms)
+✔ a coder with a tracker gets its own task tools and no task_reassign (1.922068ms)
+✔ launch only when the business's launch block names the instance as launcher (1.571719ms)
+✔ an action outside the instance's authority has no tool (2.458934ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (10.132477ms)
+✔ task.create needs a recorded human request and a requirement id (42.270988ms)
+✔ only labels named in the business file can be written (28.695884ms)
+✔ task.schedule sets and clears a due date and relations (35.181345ms)
+✔ assign and reassign move the role bots and record task.assigned (55.663835ms)
+✔ task.update.assigned is for the assignee and records task.state (55.137406ms)
+✔ a wrong expected digest records task.conflict and writes nothing (21.171762ms)
+✔ a cross-role verb needs a resolved decision, used once (27.361206ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (28.545856ms)
+✔ a lost answer is settled by a re-read and never retried (30.01709ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (37.911184ms)
+✔ a task the sync bot cannot read refuses and records nothing (18.605202ms)
+✔ verbs and polls for one business run one at a time (93.947085ms)
+✔ a due date with milliseconds is written to the second (72.13725ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (45.828384ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (24.061498ms)
+✔ a create whose final read fails succeeds and records task.created (47.741873ms)
+✔ an edit between the last write and the final read shows as external on the next poll (43.688712ms)
+✔ task.created is recorded when a later label write fails (19.256173ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 56
+ℹ fail 5
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 843.294318
+
+✖ failing tests:
+
+test at packages/business/tests/resolve.test.mjs:69:1
+✖ limits narrow the definition and never widen it (4.478519ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:87:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:91:1
+✖ role.launch stays within-role only for the instance the launch block names (5.021168ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:98:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:112:1
+✖ limits.authority without role.launch leaves the launcher with no launch block (1.627639ms)
+  AssertionError [ERR_ASSERTION]: agent
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:119:12)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: false,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:124:1
+✖ limits.authority narrows cross-role actions too (1.276774ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:132:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:136:1
+✖ classify (1.320238ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:141:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.diff b/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.diff
new file mode 100644
index 000000000..c187eaa7a
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-candidate/orig	2026-10-11 18:07:23.087807382 -0500
++++ packages/business/src/resolve.mjs	2026-10-11 18:14:57.707967161 -0500
+@@ -65,7 +65,7 @@
+   if (vars["limits.network"]) network = narrowerNetwork(network, vars["limits.network"]);
+   if (vars["limits.authority"]) {
+     withinRole = withinRole.filter((a) => vars["limits.authority"].includes(a));
+-    crossRole = crossRole.filter((a) => vars["limits.authority"].includes(a));
++    crossRole = [...crossRole];
+   }
+   // role.launch needs the business file's launch block naming this
+   // instance (addendum A section 8), and the verb must survive
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.log b/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.log
new file mode 100644
index 000000000..37e409e5b
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/resolve-cross-unnarrowed.log
@@ -0,0 +1,129 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (33.103276ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (28.078672ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (23.400199ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (16.425704ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (12.498957ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (12.625172ms)
+✔ task action subjects and linked decision trail are complete and ordered (23.3379ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (8.693012ms)
+✔ business isolation includes inherited object names and cross-business message references (19.263813ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (21.458088ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (9.875642ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (17.784604ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (11.20225ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (18.979522ms)
+✔ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (16.746697ms)
+✔ systemVars flattens the validated config (3.032988ms)
+✔ precedence: system, business, project, project role, agent (7.573478ms)
+✔ limits narrow the definition and never widen it (3.723962ms)
+✔ role.launch stays within-role only for the instance the launch block names (7.177622ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (3.083517ms)
+✖ limits.authority narrows cross-role actions too (2.477071ms)
+✔ classify (1.655321ms)
+✔ the record carries what the broker and launcher need (1.756283ms)
+✔ digest: key order doesn't matter, any value change does (8.241698ms)
+✔ refusals (3.408223ms)
+✔ the four shipped version 2 roles load (4.389377ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.557139ms)
+✔ shipped authority follows the note's table (0.924885ms)
+✔ version 1 files keep loading with no authority (1.425297ms)
+✔ the conductor policy isn't a role (0.314017ms)
+✔ a missing role file is exit 4, a symbolic link too (0.622313ms)
+✔ version 2 refusals (2.290655ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (3.899517ms)
+✔ credentials: Gitea scopes (1.46446ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.844948ms)
+✔ credentials: services (0.874648ms)
+✔ contract: a non-empty regular Markdown file beside the role file (0.942239ms)
+✔ the PM gets launch, its task verbs and the reads (10.058394ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (3.002875ms)
+✔ a coder with a tracker gets its own task tools and no task_reassign (2.268428ms)
+✔ launch only when the business's launch block names the instance as launcher (1.651138ms)
+✖ an action outside the instance's authority has no tool (4.595551ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (10.307219ms)
+✔ task.create needs a recorded human request and a requirement id (50.783873ms)
+✔ only labels named in the business file can be written (27.907976ms)
+✔ task.schedule sets and clears a due date and relations (33.79153ms)
+✔ assign and reassign move the role bots and record task.assigned (61.7935ms)
+✔ task.update.assigned is for the assignee and records task.state (54.786754ms)
+✔ a wrong expected digest records task.conflict and writes nothing (21.537052ms)
+✔ a cross-role verb needs a resolved decision, used once (25.702853ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (25.182622ms)
+✔ a lost answer is settled by a re-read and never retried (28.600271ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (39.54527ms)
+✔ a task the sync bot cannot read refuses and records nothing (16.765549ms)
+✔ verbs and polls for one business run one at a time (89.463681ms)
+✔ a due date with milliseconds is written to the second (66.833039ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (64.467321ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (23.033961ms)
+✔ a create whose final read fails succeeds and records task.created (36.136121ms)
+✔ an edit between the last write and the final read shows as external on the next poll (39.918485ms)
+✔ task.created is recorded when a later label write fails (13.988664ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 59
+ℹ fail 2
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 852.228915
+
+✖ failing tests:
+
+test at packages/business/tests/resolve.test.mjs:124:1
+✖ limits.authority narrows cross-role actions too (2.477071ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    {
+      crossRole: [
+        'task.priority.change',
+  +     'task.scope.change'
+      ],
+      withinRole: [
+        'task.create',
+        'task.close'
+      ]
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:130:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: { withinRole: [ 'task.create', 'task.close' ], crossRole: [ 'task.priority.change', 'task.scope.change' ] },
+    expected: { withinRole: [ 'task.create', 'task.close' ], crossRole: [ 'task.priority.change' ] },
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
+
+test at packages/harness/tests/tools.test.mjs:52:1
+✖ an action outside the instance's authority has no tool (4.595551ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+  +   'task_scope_change',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+      'trail',
+      'list_tasks'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/harness/tests/tools.test.mjs:54:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'task_scope_change', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.diff b/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.diff
new file mode 100644
index 000000000..6ff8920ed
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-candidate/orig	2026-10-11 18:07:39.870939784 -0500
++++ roles/coder.json	2026-10-11 18:14:55.818223939 -0500
+@@ -7,7 +7,7 @@
+   "network": "api-only",
+   "authority": {
+     "withinRole": ["task.update.assigned", "git.push.working", "review.request", "message.send"],
+-    "crossRole": ["task.scope.change"]
++    "crossRole": ["task.reassign", "task.scope.change"]
+   },
+   "credentials": [
+     {
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.log b/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.log
new file mode 100644
index 000000000..17f77e118
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/role-keeps-reassign.log
@@ -0,0 +1,169 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (25.610404ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (26.850363ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (23.84385ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (12.839692ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (14.696717ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (13.166518ms)
+✔ task action subjects and linked decision trail are complete and ordered (19.883322ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (10.292636ms)
+✔ business isolation includes inherited object names and cross-business message references (17.143026ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (20.432048ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (9.413307ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (17.112756ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (11.689255ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (22.916955ms)
+✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (25.762402ms)
+✔ systemVars flattens the validated config (3.032631ms)
+✔ precedence: system, business, project, project role, agent (8.016805ms)
+✔ limits narrow the definition and never widen it (3.808311ms)
+✔ role.launch stays within-role only for the instance the launch block names (9.313641ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (4.009638ms)
+✔ limits.authority narrows cross-role actions too (1.829767ms)
+✖ classify (2.894526ms)
+✔ the record carries what the broker and launcher need (1.969378ms)
+✔ digest: key order doesn't matter, any value change does (9.350735ms)
+✔ refusals (2.563428ms)
+✔ the four shipped version 2 roles load (4.282478ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.481512ms)
+✖ shipped authority follows the note's table (1.731521ms)
+✔ version 1 files keep loading with no authority (1.117387ms)
+✔ the conductor policy isn't a role (0.251829ms)
+✔ a missing role file is exit 4, a symbolic link too (0.421436ms)
+✔ version 2 refusals (1.742865ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (2.625104ms)
+✔ credentials: Gitea scopes (2.325317ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (2.201409ms)
+✔ credentials: services (1.168802ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.325241ms)
+✔ the PM gets launch, its task verbs and the reads (8.84067ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (2.555972ms)
+✖ a coder with a tracker gets its own task tools and no task_reassign (3.580158ms)
+✔ launch only when the business's launch block names the instance as launcher (1.736644ms)
+✔ an action outside the instance's authority has no tool (3.187475ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (10.303208ms)
+✔ task.create needs a recorded human request and a requirement id (47.944003ms)
+✔ only labels named in the business file can be written (26.77445ms)
+✔ task.schedule sets and clears a due date and relations (33.848522ms)
+✔ assign and reassign move the role bots and record task.assigned (61.702775ms)
+✔ task.update.assigned is for the assignee and records task.state (68.22306ms)
+✔ a wrong expected digest records task.conflict and writes nothing (26.375322ms)
+✔ a cross-role verb needs a resolved decision, used once (45.781054ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (58.667462ms)
+✔ a lost answer is settled by a re-read and never retried (47.540574ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (48.995082ms)
+✔ a task the sync bot cannot read refuses and records nothing (21.496242ms)
+✔ verbs and polls for one business run one at a time (115.361345ms)
+✔ a due date with milliseconds is written to the second (77.442629ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (44.709765ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (20.37596ms)
+✔ a create whose final read fails succeeds and records task.created (36.039383ms)
+✔ an edit between the last write and the final read shows as external on the next poll (40.908102ms)
+✔ task.created is recorded when a later label write fails (16.705741ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 57
+ℹ fail 4
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 962.043519
+
+✖ failing tests:
+
+test at packages/bus/tests/broker.test.mjs:361:1
+✖ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (25.762402ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  + actual - expected
+  
+  + 'cross-role'
+  - 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/bus/tests/broker.test.mjs:374:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross-role',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/resolve.test.mjs:136:1
+✖ classify (2.894526ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly equal:
+  
+  'cross' !== 'gated'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/resolve.test.mjs:141:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: 'cross',
+    expected: 'gated',
+    operator: 'strictEqual',
+    diff: 'simple'
+  }
+
+test at packages/business/tests/role.test.mjs:61:1
+✖ shipped authority follows the note's table (1.731521ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+  +   'task.reassign',
+      'task.scope.change'
+    ]
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/business/tests/role.test.mjs:65:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'task.reassign', 'task.scope.change' ],
+    expected: [ 'task.scope.change' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
+
+test at packages/harness/tests/tools.test.mjs:38:1
+✖ a coder with a tracker gets its own task tools and no task_reassign (3.580158ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+      'send_message',
+  +   'task_reassign',
+      'task_scope_change',
+      'task_update_assigned',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/harness/tests/tools.test.mjs:40:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'send_message', 'task_reassign', 'task_scope_change', 'task_update_assigned', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'send_message', 'task_scope_change', 'task_update_assigned', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.diff b/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.diff
new file mode 100644
index 000000000..a18790b8a
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.diff
@@ -0,0 +1,11 @@
+--- /mnt/storage/scratch/rocko-r73-out/mutants-candidate/orig	2026-10-11 18:07:23.096112289 -0500
++++ packages/harness/src/tools.mjs	2026-10-11 18:14:59.504647794 -0500
+@@ -89,7 +89,7 @@
+   }
+   if (tracker) {
+     for (const [action, [name, description, properties, required]] of Object.entries(TASK_TOOLS)) {
+-      if (held.has(action)) add(name, action, action, description, obj(properties, required));
++      add(name, action, action, description, obj(properties, required));
+     }
+   }
+   add("raise_decision", "decision.raise", null, "Ask for a decision before a gated or cross-role action, or when you need a choice made. The answer arrives as a message.", obj({
diff --git a/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.log b/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.log
new file mode 100644
index 000000000..e8eb463fb
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/mutants-candidate/tools-all-task-verbs.log
@@ -0,0 +1,142 @@
+✔ launch identity is stamped, payload identity is refused and stale holder cannot send (36.639909ms)
+✔ decision classes route from policy; gated resolution is human-only, choice and target must match (33.673177ms)
+✔ claim exclusion, holder release, gated revoke and rerouting to a new holder are atomic (32.106496ms)
+✔ launch events require a human CLI capability; generic emit cannot forge authority events (18.481664ms)
+✔ within-role decisions close atomically and invalid options or blocking omissions refuse (16.132294ms)
+✔ observer capabilities read human inbox but cannot mutate or forge launch identity (16.873703ms)
+✔ task action subjects and linked decision trail are complete and ordered (24.302336ms)
+✔ launch binding is durable and reconnecting requires the identical trusted record (10.520395ms)
+✔ business isolation includes inherited object names and cross-business message references (17.798456ms)
+✔ authority never transfers between action, run, target, unresolved or replaced role holder (25.213949ms)
+✔ task projection uses schema current view, skipping earlier and equal-start polls (13.067758ms)
+✔ revocation permanently bars the old run from reclaiming first, including after broker restart (20.466512ms)
+✔ empty message references refuse before storage; refusal-evidence failure stays a typed error (12.744716ms)
+✔ both arbiters require human resolution when their cross-role route is themselves (19.443375ms)
+✔ the shipped coder role raises task.reassign as gated to the human; task.scope.change stays cross-role (17.797876ms)
+✔ systemVars flattens the validated config (3.526166ms)
+✔ precedence: system, business, project, project role, agent (8.517006ms)
+✔ limits narrow the definition and never widen it (4.025856ms)
+✔ role.launch stays within-role only for the instance the launch block names (7.700117ms)
+✔ limits.authority without role.launch leaves the launcher with no launch block (5.183853ms)
+✔ limits.authority narrows cross-role actions too (1.75868ms)
+✔ classify (2.147984ms)
+✔ the record carries what the broker and launcher need (1.566134ms)
+✔ digest: key order doesn't matter, any value change does (10.446916ms)
+✔ refusals (3.104062ms)
+✔ the four shipped version 2 roles load (5.416941ms)
+✔ shipped role scopes match addendum B section 2 and the SR runbook (1.42006ms)
+✔ shipped authority follows the note's table (0.78467ms)
+✔ version 1 files keep loading with no authority (1.197489ms)
+✔ the conductor policy isn't a role (0.278589ms)
+✔ a missing role file is exit 4, a symbolic link too (0.489048ms)
+✔ version 2 refusals (1.578833ms)
+✔ authority: closed vocabulary, no gated-only action, no overlap (2.399299ms)
+✔ credentials: Gitea scopes (1.394806ms)
+✔ credentials: Vikunja scopes are a group-to-verbs map from the grantable list (1.771779ms)
+✔ credentials: services (1.393594ms)
+✔ contract: a non-empty regular Markdown file beside the role file (1.18444ms)
+✔ the PM gets launch, its task verbs and the reads (9.197271ms)
+✔ a coder gets no launch, no resolve_decision, and no task tools without a tracker (3.379526ms)
+✖ a coder with a tracker gets its own task tools and no task_reassign (5.596139ms)
+✔ launch only when the business's launch block names the instance as launcher (3.973547ms)
+✖ an action outside the instance's authority has no tool (3.762374ms)
+✔ callTool: one JSON line out, the result back, a refusal rejects (15.066181ms)
+✔ task.create needs a recorded human request and a requirement id (62.733557ms)
+✔ only labels named in the business file can be written (34.096604ms)
+✔ task.schedule sets and clears a due date and relations (44.787245ms)
+✔ assign and reassign move the role bots and record task.assigned (66.206408ms)
+✔ task.update.assigned is for the assignee and records task.state (57.83302ms)
+✔ a wrong expected digest records task.conflict and writes nothing (26.848484ms)
+✔ a cross-role verb needs a resolved decision, used once (26.339943ms)
+✔ task.close needs a verdict; after it every verb refuses with task-done (30.531726ms)
+✔ a lost answer is settled by a re-read and never retried (28.266487ms)
+✔ a create whose answer is lost is reported uncertain, and the poll finds the task (39.805487ms)
+✔ a task the sync bot cannot read refuses and records nothing (21.989204ms)
+✔ verbs and polls for one business run one at a time (99.205086ms)
+✔ a due date with milliseconds is written to the second (75.658928ms)
+✔ every write landed and the final read failed: the verb succeeds and records what it wrote (39.315806ms)
+✔ some writes landed and the final read failed: write-uncertain, and nothing is recorded (19.951694ms)
+✔ a create whose final read fails succeeds and records task.created (42.258201ms)
+✔ an edit between the last write and the final read shows as external on the next poll (97.83506ms)
+✔ task.created is recorded when a later label write fails (26.007172ms)
+ℹ tests 61
+ℹ suites 0
+ℹ pass 59
+ℹ fail 2
+ℹ cancelled 0
+ℹ skipped 0
+ℹ todo 0
+ℹ duration_ms 987.057475
+
+✖ failing tests:
+
+test at packages/harness/tests/tools.test.mjs:38:1
+✖ a coder with a tracker gets its own task tools and no task_reassign (5.596139ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+      'send_message',
+  +   'task_create',
+  +   'task_assign',
+  +   'task_reassign',
+  +   'task_schedule',
+  +   'task_priority_change',
+      'task_scope_change',
+      'task_update_assigned',
+  +   'task_close',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+      'trail',
+      'list_tasks'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/harness/tests/tools.test.mjs:40:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'send_message', 'task_create', 'task_assign', 'task_reassign', 'task_schedule', 'task_priority_change', 'task_scope_change', 'task_update_assigned', 'task_close', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'send_message', 'task_scope_change', 'task_update_assigned', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
+
+test at packages/harness/tests/tools.test.mjs:52:1
+✖ an action outside the instance's authority has no tool (3.762374ms)
+  AssertionError [ERR_ASSERTION]: Expected values to be strictly deep-equal:
+  + actual - expected
+  
+    [
+  +   'task_create',
+  +   'task_assign',
+  +   'task_reassign',
+  +   'task_schedule',
+  +   'task_priority_change',
+  +   'task_scope_change',
+  +   'task_update_assigned',
+  +   'task_close',
+      'raise_decision',
+      'list_agents',
+      'inbox',
+      'trail',
+      'list_tasks'
+  
+      at TestContext.<anonymous> (file:///mnt/storage/scratch/rocko-r73/packages/harness/tests/tools.test.mjs:54:10)
+      at Test.runInAsyncScope (node:async_hooks:226:14)
+      at Test.run (node:internal/test_runner/test:1402:25)
+      at Test.processPendingSubtests (node:internal/test_runner/test:974:18)
+      at Test.postRun (node:internal/test_runner/test:1542:19)
+      at Test.run (node:internal/test_runner/test:1467:12)
+      at async Test.processPendingSubtests (node:internal/test_runner/test:974:7) {
+    generatedMessage: true,
+    code: 'ERR_ASSERTION',
+    actual: [ 'task_create', 'task_assign', 'task_reassign', 'task_schedule', 'task_priority_change', 'task_scope_change', 'task_update_assigned', 'task_close', 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    expected: [ 'raise_decision', 'list_agents', 'inbox', 'trail', 'list_tasks' ],
+    operator: 'deepStrictEqual',
+    diff: 'simple'
+  }
diff --git a/agents/rocko/work/queue-73/out/old-narrow-test-under-mutant.txt b/agents/rocko/work/queue-73/out/old-narrow-test-under-mutant.txt
new file mode 100644
index 000000000..5db6ce228
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/old-narrow-test-under-mutant.txt
@@ -0,0 +1,2 @@
+ℹ pass 1
+ℹ fail 0
diff --git a/agents/rocko/work/queue-73/out/pre.md5 b/agents/rocko/work/queue-73/out/pre.md5
new file mode 100644
index 000000000..941a3e9c3
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/pre.md5
@@ -0,0 +1,4 @@
+900b080b9d61e9aa00fd7e6aaf04d84d  roles/coder.json
+3773d12359d85a4ec93bcdd2c2b96101  packages/bus/src/broker.mjs
+0b70de33a0391963899fa3d98613bf40  packages/business/src/resolve.mjs
+e26512bb2646ce69ef9bd127ab6c6b96  packages/harness/src/tools.mjs
diff --git a/agents/rocko/work/queue-73/out/probe-base.txt b/agents/rocko/work/queue-73/out/probe-base.txt
new file mode 100644
index 000000000..2f0d7ca61
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/probe-base.txt
@@ -0,0 +1,3 @@
+task.reassign class cross-role route_to pm
+task.scope.change class cross-role route_to pm
+coder tools with tracker: send_message,task_reassign,task_scope_change,task_update_assigned,raise_decision,list_agents,inbox,trail,list_tasks
diff --git a/agents/rocko/work/queue-73/out/probe-candidate.txt b/agents/rocko/work/queue-73/out/probe-candidate.txt
new file mode 100644
index 000000000..6486006d0
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/probe-candidate.txt
@@ -0,0 +1,3 @@
+task.reassign class gated route_to human
+task.scope.change class cross-role route_to pm
+coder tools with tracker: send_message,task_scope_change,task_update_assigned,raise_decision,list_agents,inbox,trail,list_tasks
diff --git a/agents/rocko/work/queue-73/out/probe.mjs b/agents/rocko/work/queue-73/out/probe.mjs
new file mode 100644
index 000000000..77f1bb766
--- /dev/null
+++ b/agents/rocko/work/queue-73/out/probe.mjs
@@ -0,0 +1,27 @@
+// Probe: what the broker and harness do today with the shipped coder role.
+import { mkdtempSync, rmSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+const tree = process.argv[2];
+const { Store } = await import(join(tree, 'packages/bus/src/store.mjs'));
+const { Broker } = await import(join(tree, 'packages/bus/src/broker.mjs'));
+const { busBusiness } = await import(join(tree, 'packages/bus/src/business.mjs'));
+const { validateBusinessDocument, resolveInstance } = await import(join(tree, 'packages/business/src/index.mjs'));
+const { businessDoc, REPO_ROLES, systemFor } = await import(join(tree, 'packages/business/tests/helpers.mjs'));
+const { typedTools } = await import(join(tree, 'packages/harness/src/tools.mjs'));
+const root = mkdtempSync(join(tmpdir(), 'r73-probe-'));
+try {
+  const business = validateBusinessDocument(businessDoc(root, 'demo'), join(root, 'businesses', 'demo.json'), { rolesDir: REPO_ROLES });
+  const resolved = Object.fromEntries(Object.keys(business.roles).map((i) => [i, resolveInstance({ system: systemFor(root), business, instance: i })]));
+  const store = new Store(root);
+  const b = new Broker({ store, businesses: { demo: busBusiness(business, resolved) } });
+  const c = b.bindLaunch({ business: 'demo', role: 'coder', run: 'coder-run', harness: 'pi', address: 'coder-run' });
+  b.request(c, { verb: 'role.claim', args: {} });
+  const options = [{ key: 'yes', text: 'Allow' }, { key: 'no', text: 'Decline' }];
+  for (const action of ['task.reassign', 'task.scope.change']) {
+    const d = b.request(c, { verb: 'decision.raise', args: { action, question: 'Allow?', options, recommendation: 'no', blocking: false } });
+    console.log(action, 'class', d.class, 'route_to', d.route_to);
+  }
+  console.log('coder tools with tracker:', typedTools({ resolved: resolved.coder, tracker: true }).map((t) => t.name).join(','));
+  store.close();
+} finally { rmSync(root, { recursive: true, force: true }); }
diff --git a/agents/rocko/work/queue-73/packet-manifest.sha256 b/agents/rocko/work/queue-73/packet-manifest.sha256
new file mode 100644
index 000000000..6c081668e
--- /dev/null
+++ b/agents/rocko/work/queue-73/packet-manifest.sha256
@@ -0,0 +1,53 @@
+a101b9dc7a08d79018dd8c6ac77dcf451901b0b8910de1a792367cf1f4a35b1a  ./base.txt
+57e963e58c1f17fb0cb8cbf008ba5e305fc4555d4f0696723c85edc7871f8256  ./BUILD.md
+63db27a4bd80eec09da09c1a3b7de7977d096447de313b2f42de5c60efbeeb60  ./build.patch
+bf4e5ed646301d06617004770aaaf9599e743f1e00d226f8a9e0c2450f12710d  ./candidate-manifest.sha256
+516fbe81e33a9d44351758deca7d18d580a7a1d640510ec2941906ed277bdadd  ./files.txt
+e3300acd1f85f7c7c495dc2228ec904c0af992f9ea09693adcf3e5b1121d5f83  ./mutants.sh
+f8b54deb8f32f8c420ed5a503d4623e89bee762ff2af5ce74185b1fc9809024b  ./out/gate/business.log
+b364089f6df19de5086c1c15258696fd9a719cfcbb672ec5d504a065fc01178e  ./out/gate/bus.log
+bbbf01cda98f6710f35e48a0ed93277b98b010bfa27d7bad1c13d46d40ea4bd4  ./out/gate/cli.log
+cb7eb33230c15549e0d6cccf84125db76c4f08a420da585560d4d2cea3fc3935  ./out/gate/end.txt
+0eb63fc07e1e77a362a26513b46be8fd77547c033082b4ce969c4819cd4b9898  ./out/gate/harness.log
+84dfcd7429af68d3c9e1cff2044dc3c8a33524931333f6b6c7842107c1809058  ./out/gate/seat.log
+bb11ccdba6113bd6441edffadbfc31889ad31a8ae5a1b7730efd76479f70a0f5  ./out/gate.sh
+f594c4c232b18e1c5e943a5bbe0776c34768a850923ee6d796f1bcc6cd81ca3b  ./out/gate/start.txt
+9c18698db3974473d3f5c4adf65d959050c9dcc356daff8a307274fb7274be53  ./out/gate/summary.txt
+afefb840318fad067fd0080e65559080c09e2e58d3bd6bdd95ce1133cbe4954d  ./out/gate/tasks.log
+f91fc95a173339137f981e24fbf630acfde957e32c3cd71e28304532083c6a77  ./out/gate/test-auth.log
+e1206366f07f3d9029530dcdcb55c0282d3f052adc60fc7fd51642d5d4ec3600  ./out/gate/test-conductor.log
+52d211444489ea729dafc3c9745d4474d9dfd50f41055cbad792424197f04073  ./out/gate/test-config.log
+7d5016ded1994048642f8b9ae845e52e8af4a069ba05c056c8f580a4589bacbe  ./out/gate/test-discord.log
+1ae08907ecdf1fa939b6f051b22d28462086a9d9f0f76c2d35b9ba2b21132e27  ./out/gate/test-extension-package.log
+83dfaa21c4e4ad941088efab3f69e87ebd2f6bd3cc72ede938e21dc06da8a34f  ./out/gate/test-foundation.log
+b413b13dda4ccebcf63dd3d5958e0d6cdfcacc124a8f478b712bccc8c69ffd52  ./out/gate/test-queue.log
+fa832794ee35025fb4559d58d01a270f6bff2eb65e863e9075b797f47fa68545  ./out/gate/test-release.log
+ccea37f4ffde1f160dcd1a1864988083a645d974e892065c956b338d0f8c853a  ./out/gate/test-task.log
+025ce7565cacda5a0e84d23a87ee040eeca76605178a9da0344f1de85ed51284  ./out/gate/webui.log
+405f17ea5490a22cd8175ad3d70383a0c7df6a9e9f4b8a0a773117bf97a1a667  ./out/mutants-base/baseline.log
+c473c06ab442b60be7352c16b42de590c32d981f666b32c8bca8a7ca57fc37eb  ./out/mutants-base/broker-unlisted-cross.diff
+92c8cf650edd7ec6d7fdbe456b28e6d2cb1c99b5e9bf772a4468db91d2e2e5ff  ./out/mutants-base/broker-unlisted-cross.log
+98b872bc68686436afa6024a4a86466dfaefa0eebd24eef583248dab2fdd52a7  ./out/mutants-base/classify-unlisted-cross.diff
+5b48e8dc706edc1df9a93471aca575efc4cadca283c5acfe0dc3c6d332d74b0c  ./out/mutants-base/classify-unlisted-cross.log
+149190c6f27c506716dbb90df1edd4872ad44c58b78cc77f4a0cf40e9d5e2e8b  ./out/mutants-base/resolve-cross-unnarrowed.diff
+5edaee8eba9467a5f9c78fcf55c13dff593412e20d44f2a85638035775335adb  ./out/mutants-base/resolve-cross-unnarrowed.log
+4fbcfbcd1a3b2307e1e3fc0ff4bcf1ffa3e90e6c600257a99c93aeeb9719b722  ./out/mutants-base/tools-all-task-verbs.diff
+6c92f1d4b6e77ac5a2cb9c5a744a0c49d068b6fe0367716fd2529e553c70a8c3  ./out/mutants-base/tools-all-task-verbs.log
+4ecd1ba8a8515846f15fe3af3ade57b5e04abb758f05d50a52e3c54f25c971f1  ./out/mutants-base.txt
+a410b4ab2f7718bf7dc424f4fa3394ab9f59f70e396acd4dbd54a5933d3b2e53  ./out/mutants-candidate/baseline.log
+cc6be3793b2e5e77dde9adaeda87d73153c2e5286a6582b26f8ca54699a7a1a9  ./out/mutants-candidate/broker-unlisted-cross.diff
+f8d26304746eeeec72dd4b7f0e4bfa34a9b393156dd89ae0cc4df338bfe27774  ./out/mutants-candidate/broker-unlisted-cross.log
+61e5bf405c309604799f1bfa3a87c04c5900673a55891c80a8859a8357985212  ./out/mutants-candidate/classify-unlisted-cross.diff
+500f3837c9bea02e2228a8ade0f143649225ff6854d3d9468c9b4dddb50ffd92  ./out/mutants-candidate/classify-unlisted-cross.log
+2bb9f2a49ec99852ce2c7ed19e6f01fc061c0f1a1a0aaa9e4a8f7469c41efe29  ./out/mutants-candidate/resolve-cross-unnarrowed.diff
+9a7aadcaa01dc7b92d4180f4cd6a9ae266cfbcf4f7b49c4854b2dc13f06e0986  ./out/mutants-candidate/resolve-cross-unnarrowed.log
+ca8f11dedc9de9bda0ea127548680b9c8cbdd0d8a9bf9352b0a42ff37f686f0c  ./out/mutants-candidate/role-keeps-reassign.diff
+96d653c7368821d451fae7cf57ea2bf005b4ab542e1a3d1c27294fe832b58c2f  ./out/mutants-candidate/role-keeps-reassign.log
+c3d83cbbb17b3bd788089610b3ee440f2c7e8124a4a8032e4f475e48d030ca55  ./out/mutants-candidate/tools-all-task-verbs.diff
+75a35b18ea42f629fb45e620703c6647eff0e3efaf6968739e5d66ef718e40c7  ./out/mutants-candidate/tools-all-task-verbs.log
+60d077be2cfad3dc9c365a0701b535b48cbaea5a87def47832992aa1f4138481  ./out/mutants-candidate.txt
+719ccf9999d00adaaab22be5247a231e84d1e3077f9b8687c09caf4ab7bf63aa  ./out/old-narrow-test-under-mutant.txt
+82479aa50e513a3c764d3364298ddb1a3d148c38e32942de3f0f91c7c8df874b  ./out/pre.md5
+42475e9e7238f9daa54b54c1d37512ca172116918c48150ec6abc86047d7aa62  ./out/probe-base.txt
+ae7e5be038f5b5db667bb7f4d08a00e6b1d5827c8024857ac642a8b1f8128452  ./out/probe-candidate.txt
+3d44b91cd0e2b40e935812b35c41ef40df5e267e517f51af217ab38feb8507e1  ./out/probe.mjs
diff --git a/docs/SESSIONS.md b/docs/SESSIONS.md
index 914e4c112..38922b831 100644
--- a/docs/SESSIONS.md
+++ b/docs/SESSIONS.md
@@ -687,3 +687,5 @@ are never rewritten or removed; corrections are new entries.
 - 2026-10-11T23:06:07Z, Sage (lead): row 72 (#1566) round 1. I posted the review request as jarvis for Rocko, comment 27454 (revs 527-528), and committed Rocko's packet (candidate e6531a53) with their revs 524-526 and their BUILD-LOG and SESSIONS appends. Reviewer: Filbert. Also row 74 goes to round 2 for Filbert's and Darkwing's sign-in findings (notes revs 521, 523); and 166cba45, my queue-commit with subject 'x', was rewritten locally with the same tree as ebd90554 before any push.
 2026-10-11T23:12:50Z | filbert | row 72 (#1566) round 1 review, Claude pin tests | approve, comment 27460, rev 529 in 0a4b7a4e; Rocko's 6 mutants killed (all survive on base), 7 of my 8 extra killed; N1 the default path skips no test of the checks (non-exec claude, PATH dir with a space), follow-up; cli 113, harness 58 green
 2026-10-11T23:14:17Z | dewey | row 58 (#1547) round 1 review, task priority and requirement | approve, comment 27461, rev 530 in 05d9863c; 7 of my 10 mutants killed, 2 equivalent; N1 priority 0 'unset' isn't pinned in the view (priority-zero-none survives), N2 Filbert's N1 and N2 into the same follow-up; gate green on ae428a54, discord 178/0
+2026-10-11T23:17:20Z | rocko | row 73 (#1567) round 1 candidate, drop the coder task.reassign grant | in-review rev 533, request failed rev 534 (no credential, Sage to post as jarvis); packet agents/rocko/work/queue-73 uncommitted, candidate manifest bf4e5ed6, packet 41ffcd5c; 5/5 mutants killed (broker-unlisted-cross survives base); gate green; reviewer Dewey
+- 2026-10-11T23:20:28Z, Sage (lead): row 73 (#1567) round 1. I posted the review request as jarvis for Rocko, comment 27467 (revs 537-538), and committed Rocko's packet (candidate bf4e5ed6, packet 41ffcd5c) with their BUILD-LOG and SESSIONS appends. Reviewer: Dewey. The three stale places Rocko named (business README example, tasks README lines 193-196, tasks live/run.mjs line 32) go to row 79 (#1573, rev 539), owner Rocko after 73, reviewer Dewey, so round 1 isn't restarted.