mandrel 2.87.0 → 2.89.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (170) hide show
  1. package/.agents/README.md +29 -0
  2. package/.agents/agents/auditor.md +4 -3
  3. package/.agents/docs/agentrc-reference.json +5 -2
  4. package/.agents/docs/configuration.md +14 -8
  5. package/.agents/docs/workflows.md +1 -1
  6. package/.agents/schemas/agentrc.schema.json +24 -5
  7. package/.agents/schemas/story-deliver-terminal.schema.json +1 -1
  8. package/.agents/schemas/story-progress.schema.json +5 -0
  9. package/.agents/scripts/audit-to-stories.js +3 -2
  10. package/.agents/scripts/boot-sweep.js +3 -2
  11. package/.agents/scripts/check-arch-cycles.js +19 -41
  12. package/.agents/scripts/check-context-budget.js +16 -44
  13. package/.agents/scripts/check-cyclomatic.js +13 -37
  14. package/.agents/scripts/check-dead-exports.js +16 -46
  15. package/.agents/scripts/clean-worktrees.js +1 -6
  16. package/.agents/scripts/coverage-capture.js +32 -49
  17. package/.agents/scripts/deliver-run.js +11 -0
  18. package/.agents/scripts/diagnose-friction.js +3 -3
  19. package/.agents/scripts/lib/audit-baselines/gate-surface.js +24 -11
  20. package/.agents/scripts/lib/audit-baselines/read.js +4 -2
  21. package/.agents/scripts/lib/audit-baselines/surface-entry.js +95 -11
  22. package/.agents/scripts/lib/baselines/coverage-updater-cli.js +6 -14
  23. package/.agents/scripts/lib/baselines/crap-preview-scan.js +3 -1
  24. package/.agents/scripts/lib/baselines/crap-updater-cli.js +38 -26
  25. package/.agents/scripts/lib/baselines/duplication-scanner.js +20 -1
  26. package/.agents/scripts/lib/baselines/env-overrides.js +39 -34
  27. package/.agents/scripts/lib/baselines/kernel.js +3 -2
  28. package/.agents/scripts/lib/baselines/kinds/crap.js +50 -53
  29. package/.agents/scripts/lib/baselines/preview-gates.js +5 -1
  30. package/.agents/scripts/lib/baselines/refresh-service.js +65 -22
  31. package/.agents/scripts/lib/baselines/writer.js +5 -1
  32. package/.agents/scripts/lib/bdd-runner-detect.js +7 -10
  33. package/.agents/scripts/lib/child-exec.js +14 -1
  34. package/.agents/scripts/lib/child-stream.js +122 -0
  35. package/.agents/scripts/lib/clean-temp.js +19 -13
  36. package/.agents/scripts/lib/cli-args.js +49 -0
  37. package/.agents/scripts/lib/close-validation/gates.js +40 -9
  38. package/.agents/scripts/lib/codex/skills-projection.js +1 -12
  39. package/.agents/scripts/lib/config/gates/crap.schema.js +15 -0
  40. package/.agents/scripts/lib/config/gates/mutation.schema.js +2 -2
  41. package/.agents/scripts/lib/config/gh-config-dir.js +63 -0
  42. package/.agents/scripts/lib/config/github.js +2 -0
  43. package/.agents/scripts/lib/config/paths.js +14 -3
  44. package/.agents/scripts/lib/config/quality.js +49 -1
  45. package/.agents/scripts/lib/config-resolver.js +78 -64
  46. package/.agents/scripts/lib/config-settings-schema-delivery.js +4 -3
  47. package/.agents/scripts/lib/config-settings-schema.js +7 -0
  48. package/.agents/scripts/lib/coverage-baseline.js +2 -2
  49. package/.agents/scripts/lib/coverage-capture-affected.js +2 -1
  50. package/.agents/scripts/lib/coverage-capture-fullscope.js +2 -2
  51. package/.agents/scripts/lib/coverage-capture-incremental.js +1 -1
  52. package/.agents/scripts/lib/coverage-capture-passes.js +143 -0
  53. package/.agents/scripts/lib/coverage-capture-usage.js +1 -1
  54. package/.agents/scripts/lib/coverage-capture.js +45 -15
  55. package/.agents/scripts/lib/crap-utils.js +65 -49
  56. package/.agents/scripts/lib/full-suite-lock.js +1 -10
  57. package/.agents/scripts/lib/gates/baseline-store.js +10 -0
  58. package/.agents/scripts/lib/generated/agentrc-validator.js +1 -1
  59. package/.agents/scripts/lib/gh-exec.js +45 -2
  60. package/.agents/scripts/lib/gh-transport.js +27 -0
  61. package/.agents/scripts/lib/git-utils.js +16 -9
  62. package/.agents/scripts/lib/maintainability-engine.js +27 -6
  63. package/.agents/scripts/lib/maintainability-utils.js +20 -2
  64. package/.agents/scripts/lib/orchestration/check-baselines/phases/evaluate.js +4 -1
  65. package/.agents/scripts/lib/orchestration/evidence-stream.js +186 -0
  66. package/.agents/scripts/lib/orchestration/file-assumptions.js +2 -19
  67. package/.agents/scripts/lib/orchestration/git-cleanup/phases/branches-reap.js +3 -7
  68. package/.agents/scripts/lib/orchestration/git-cleanup/phases/git-probes-ff.js +34 -40
  69. package/.agents/scripts/lib/orchestration/git-cleanup/phases/git-probes.js +4 -3
  70. package/.agents/scripts/lib/orchestration/plan-persist/persist-helpers.js +2 -19
  71. package/.agents/scripts/lib/orchestration/pr-watch.js +1 -5
  72. package/.agents/scripts/lib/orchestration/project-meta-cache.js +17 -4
  73. package/.agents/scripts/lib/orchestration/reassert-status-column.js +1 -8
  74. package/.agents/scripts/lib/orchestration/resolve-stories.js +1 -0
  75. package/.agents/scripts/lib/orchestration/review-providers/native.js +2 -2
  76. package/.agents/scripts/lib/orchestration/single-story-close/phase-timer.js +43 -0
  77. package/.agents/scripts/lib/orchestration/single-story-close/phases/confirm-merge.js +37 -13
  78. package/.agents/scripts/lib/orchestration/single-story-close/phases/finish.js +264 -0
  79. package/.agents/scripts/lib/orchestration/single-story-close/phases/open-pr.js +68 -0
  80. package/.agents/scripts/lib/orchestration/single-story-close/phases/pre-push.js +106 -0
  81. package/.agents/scripts/lib/orchestration/single-story-close/phases/reconcile-done.js +43 -35
  82. package/.agents/scripts/lib/orchestration/single-story-close/phases/review-halt.js +52 -0
  83. package/.agents/scripts/lib/orchestration/single-story-close/result.js +273 -0
  84. package/.agents/scripts/lib/orchestration/single-story-close/runner.js +68 -776
  85. package/.agents/scripts/lib/orchestration/story-close/context-budget-writeback.js +20 -9
  86. package/.agents/scripts/lib/orchestration/story-close/duplication-close-rescore.js +2 -2
  87. package/.agents/scripts/lib/orchestration/story-close/ratchet-shrink-writeback.js +20 -12
  88. package/.agents/scripts/lib/orchestration/story-handoff.js +50 -21
  89. package/.agents/scripts/lib/orchestration/story-progress.js +32 -12
  90. package/.agents/scripts/lib/orchestration/ticket-validator.js +3 -13
  91. package/.agents/scripts/lib/orchestration/verify-credit.js +4 -1
  92. package/.agents/scripts/lib/qa/resolve-selection.js +1 -8
  93. package/.agents/scripts/lib/single-story/confirm-and-land.js +33 -0
  94. package/.agents/scripts/lib/single-story-sweep/reconcile-closing.js +43 -11
  95. package/.agents/scripts/lib/single-story-sweep/sweep-lock.js +130 -30
  96. package/.agents/scripts/lib/skills/parse-skill.js +1 -4
  97. package/.agents/scripts/lib/story-body/story-body.js +12 -1
  98. package/.agents/scripts/lib/suite-admission.js +3 -11
  99. package/.agents/scripts/lib/temp-retention.js +25 -11
  100. package/.agents/scripts/lib/test-run-credit.js +2 -5
  101. package/.agents/scripts/lib/transpile.js +193 -31
  102. package/.agents/scripts/lib/util/paths.js +31 -0
  103. package/.agents/scripts/lib/util/sleep.js +15 -0
  104. package/.agents/scripts/lib/validation-evidence.js +29 -3
  105. package/.agents/scripts/lib/wave-runner/live-probe.js +24 -0
  106. package/.agents/scripts/lib/workers/crap-worker.js +23 -12
  107. package/.agents/scripts/lib/workers/maintainability-report-worker.js +18 -12
  108. package/.agents/scripts/lib/workflow-closure.js +1 -8
  109. package/.agents/scripts/notify.js +83 -90
  110. package/.agents/scripts/plan-context.js +8 -4
  111. package/.agents/scripts/providers/github/auth.js +14 -30
  112. package/.agents/scripts/providers/github/blocked-by-add.js +11 -27
  113. package/.agents/scripts/providers/github/compose.js +1 -0
  114. package/.agents/scripts/providers/github/gh-token-diagnostics.js +130 -0
  115. package/.agents/scripts/providers/github/identity.js +158 -0
  116. package/.agents/scripts/providers/github/issues.js +2 -2
  117. package/.agents/scripts/providers/github/mappers.js +16 -19
  118. package/.agents/scripts/providers/github/projects-v2-graphql.js +4 -36
  119. package/.agents/scripts/providers/github/request-helpers.js +44 -0
  120. package/.agents/scripts/providers/github/sub-issue-add.js +11 -28
  121. package/.agents/scripts/providers/github.js +10 -2
  122. package/.agents/scripts/quality-preview.js +6 -19
  123. package/.agents/scripts/single-story-confirm-merge.js +8 -20
  124. package/.agents/scripts/single-story-init.js +0 -1
  125. package/.agents/scripts/stories-wave-tick.js +60 -46
  126. package/.agents/scripts/story-handoff.js +2 -23
  127. package/.agents/scripts/story-review-compute.js +3 -22
  128. package/.agents/scripts/sync-claude-agents.js +1 -12
  129. package/.agents/scripts/sync-claude-commands.js +1 -12
  130. package/.agents/scripts/update-crap-baseline.js +11 -20
  131. package/.agents/scripts/update-ticket-state.js +44 -45
  132. package/.agents/templates/docs/audit-sweep-runbook.md +1 -1
  133. package/.agents/workflows/audit-accessibility.md +1 -1
  134. package/.agents/workflows/audit-adrs.md +1 -1
  135. package/.agents/workflows/audit-architecture.md +1 -1
  136. package/.agents/workflows/audit-baselines.md +9 -4
  137. package/.agents/workflows/audit-clean-code.md +1 -1
  138. package/.agents/workflows/audit-data-model.md +1 -1
  139. package/.agents/workflows/audit-dependencies.md +1 -1
  140. package/.agents/workflows/audit-devops.md +1 -1
  141. package/.agents/workflows/audit-documentation.md +1 -1
  142. package/.agents/workflows/audit-exceptions.md +2 -2
  143. package/.agents/workflows/audit-mobile.md +1 -1
  144. package/.agents/workflows/audit-navigability.md +1 -1
  145. package/.agents/workflows/audit-performance.md +1 -1
  146. package/.agents/workflows/audit-privacy.md +1 -1
  147. package/.agents/workflows/audit-quality.md +1 -1
  148. package/.agents/workflows/audit-security.md +1 -1
  149. package/.agents/workflows/audit-seo.md +1 -1
  150. package/.agents/workflows/audit-sre.md +1 -1
  151. package/.agents/workflows/audit-to-stories.md +9 -7
  152. package/.agents/workflows/audit-ux-ui.md +1 -1
  153. package/.agents/workflows/helpers/audit-lens-core.md +8 -5
  154. package/.agents/workflows/helpers/deliver-digest.md +4 -5
  155. package/.agents/workflows/helpers/deliver-story-reference.md +4 -4
  156. package/.agents/workflows/mandrel-deliver.md +3 -1
  157. package/README.md +6 -3
  158. package/docs/CHANGELOG.md +42 -0
  159. package/lib/cli/doctor-checks/codex.js +240 -0
  160. package/lib/cli/doctor-checks/merge-driver.js +79 -0
  161. package/lib/cli/doctor-checks/payload.js +201 -0
  162. package/lib/cli/doctor-checks/read-json.js +16 -0
  163. package/lib/cli/doctor-checks/runtime-deps.js +146 -0
  164. package/lib/cli/doctor-checks/spawn.js +18 -0
  165. package/lib/cli/doctor-checks/sync-parity.js +205 -0
  166. package/lib/cli/doctor-checks/test-credit-path.js +270 -0
  167. package/lib/cli/doctor-checks/toolchain.js +137 -0
  168. package/lib/cli/doctor-checks/version-current.js +82 -0
  169. package/lib/cli/registry.js +40 -1330
  170. package/package.json +1 -1
@@ -1,7 +1,7 @@
1
1
  // Must be the first import: fails fast before any third-party import loads.
2
2
  import './lib/runtime-deps/ensure-installed.js';
3
3
  import {
4
- buildCrapUpdaterScorer,
4
+ buildCrapRefreshOptions,
5
5
  parseCrapUpdaterArgs,
6
6
  resolveCrapUpdaterOptions,
7
7
  seatCrapBaseline,
@@ -15,10 +15,7 @@ import {
15
15
  resolveConfig,
16
16
  } from './lib/config-resolver.js';
17
17
  import { loadCoverage } from './lib/coverage-utils.js';
18
- import {
19
- resolveEscomplexVersion,
20
- resolveTsTranspilerVersion,
21
- } from './lib/crap-utils.js';
18
+ import { resolveEscomplexVersion } from './lib/crap-utils.js';
22
19
 
23
20
  import { Logger } from './lib/Logger.js';
24
21
 
@@ -42,7 +39,7 @@ const USAGE = {
42
39
  ],
43
40
  [
44
41
  '--coverage <path>',
45
- 'Read coverage from this path (default `coverage/coverage-final.json`).',
42
+ "Read coverage from this path; wins over config (default `delivery.quality.gates.crap.coveragePath`, falling back to the coverage gate's `coveragePath`, `coverage/coverage-final.json`).",
46
43
  ],
47
44
  [
48
45
  '--full-scope',
@@ -58,7 +55,7 @@ const USAGE = {
58
55
  ],
59
56
  ],
60
57
  notes: [
61
- 'Run `npm run test:coverage` first — without a coverage artifact every file is skipped.',
58
+ 'Run the CRAP capture script first — `npm run <delivery.quality.gates.crap.captureScript>`, default `npm run test:coverage` — without a coverage artifact the refresh refuses (exit 1) and leaves the baseline untouched.',
62
59
  ],
63
60
  };
64
61
 
@@ -75,21 +72,15 @@ async function main() {
75
72
  `[CRAP] Coverage source: ${options.coveragePath}${options.requireCoverage ? ' (required)' : ' (optional)'}`,
76
73
  );
77
74
 
78
- const refreshOpts = {
79
- kind: 'crap',
80
- writePath: options.absBaselinePath,
81
- epsilon: getBaselineEpsilon('crap'),
82
- scorer: buildCrapUpdaterScorer(options, { loadCoverage }),
83
- };
84
- // No flag -> scopeFiles=null + fullScope=false -> the service derives the
85
- // diff via `origin/main..HEAD` (its default baseRef/headRef).
86
- if (options.fullScope) refreshOpts.fullScope = true;
87
- else if (options.diffScopeRef) refreshOpts.baseRef = options.diffScopeRef;
88
-
89
- const result = await refreshBaseline(refreshOpts);
75
+ const result = await refreshBaseline(
76
+ buildCrapRefreshOptions(options, {
77
+ loadCoverage,
78
+ epsilon: getBaselineEpsilon('crap'),
79
+ }),
80
+ );
90
81
 
91
82
  Logger.info(
92
- `[CRAP] ✅ Baseline updated (kernelVersion=${result.envelope.kernelVersion}, escomplexVersion=${resolveEscomplexVersion()}, tsTranspilerVersion=${resolveTsTranspilerVersion()}). Wrote to ${options.absBaselinePath}.`,
83
+ `[CRAP] ✅ Baseline updated (kernelVersion=${result.envelope.kernelVersion}, escomplexVersion=${resolveEscomplexVersion()}, tsTranspilerVersion=${result.envelope.tsTranspilerVersion}). Wrote to ${options.absBaselinePath}.`,
93
84
  );
94
85
  Logger.info(
95
86
  `[CRAP] Wrote ${result.envelope.rows.length} row(s). scope=${result.scope.mode}, wrote=${result.wrote}.`,
@@ -6,7 +6,7 @@
6
6
  */
7
7
 
8
8
  import { parseArgs } from 'node:util';
9
- import { respondToHelp } from './lib/cli-usage.js';
9
+ import { runAsCli } from './lib/cli-utils.js';
10
10
  import { resolveConfig } from './lib/config-resolver.js';
11
11
  import { Logger } from './lib/Logger.js';
12
12
  import { transitionTicketState } from './lib/orchestration/ticketing.js';
@@ -31,15 +31,10 @@ const USAGE = {
31
31
  notes: ['Exactly one of --state or --remove-label is required.'],
32
32
  };
33
33
 
34
- // ── CLI Main Block ────────────────────────────────────────────────────────
35
- // cli-opt-out: re-export shim with a DEBUG_MAIN escape hatch for tests; runAsCli's strict path-equality guard would block the env-flag entry path.
36
- if (
37
- (process.argv[1]?.endsWith('update-ticket-state.js') ||
38
- process.env.DEBUG_MAIN) &&
39
- !respondToHelp(process.argv.slice(2), USAGE)
40
- ) {
34
+ /** @param {string[]} argv */
35
+ function parseTicketStateArgs(argv) {
41
36
  const { values } = parseArgs({
42
- args: process.argv.slice(2),
37
+ args: argv,
43
38
  options: {
44
39
  ticket: { type: 'string' },
45
40
  state: { type: 'string' },
@@ -47,11 +42,9 @@ if (
47
42
  },
48
43
  strict: false,
49
44
  });
50
-
51
45
  const ticketId = Number.parseInt(values.ticket, 10);
52
46
  const state = values.state;
53
47
  const removeLabel = values['remove-label'];
54
-
55
48
  if (Number.isNaN(ticketId) || (!state && !removeLabel)) {
56
49
  throw new Error(
57
50
  'Usage: node update-ticket-state.js ' +
@@ -59,44 +52,50 @@ if (
59
52
  '[--state <state> | --remove-label <label>]',
60
53
  );
61
54
  }
55
+ return { ticketId, state, removeLabel };
56
+ }
62
57
 
63
- (async () => {
64
- const config = resolveConfig();
65
- const provider = createProvider(config);
66
-
67
- // Label-only mutation path — no state transition. Callers that just
68
- // need to drop a single label without flipping the agent::* state.
69
- if (removeLabel && !state) {
70
- Logger.info(
71
- `[State-Sync] Removing label \`${removeLabel}\` from ticket #${ticketId}...`,
72
- );
73
- await provider.updateTicket(ticketId, {
74
- labels: { remove: [removeLabel] },
75
- });
76
- Logger.info('[State-Sync] ✅ Success');
77
- return;
78
- }
58
+ /**
59
+ * @param {{ argv?: string[], provider?: object, transition?: typeof transitionTicketState }} [opts]
60
+ */
61
+ export async function runUpdateTicketState({
62
+ argv = process.argv.slice(2),
63
+ provider,
64
+ transition = transitionTicketState,
65
+ } = {}) {
66
+ const { ticketId, state, removeLabel } = parseTicketStateArgs(argv);
67
+ const ticketing = provider ?? createProvider(resolveConfig());
79
68
 
69
+ // Label-only mutation path — no state transition. Callers that just
70
+ // need to drop a single label without flipping the agent::* state.
71
+ if (removeLabel && !state) {
80
72
  Logger.info(
81
- `[State-Sync] Transitioning ticket #${ticketId} to ${state}...`,
73
+ `[State-Sync] Removing label \`${removeLabel}\` from ticket #${ticketId}...`,
82
74
  );
83
- // No explicit cascade: `transitionTicketState` already runs it.
84
- await transitionTicketState(provider, ticketId, state);
75
+ await ticketing.updateTicket(ticketId, {
76
+ labels: { remove: [removeLabel] },
77
+ });
78
+ Logger.info('[State-Sync] ✅ Success');
79
+ return { ticketId, state: null, removedLabel: removeLabel };
80
+ }
85
81
 
86
- // Optional secondary label removal alongside the state transition
87
- // (e.g. clear `status::blocked` when transitioning back to ready).
88
- if (removeLabel) {
89
- await provider.updateTicket(ticketId, {
90
- labels: { remove: [removeLabel] },
91
- });
92
- }
82
+ Logger.info(`[State-Sync] Transitioning ticket #${ticketId} to ${state}...`);
83
+ // No explicit cascade: `transitionTicketState` already runs it.
84
+ await transition(ticketing, ticketId, state);
93
85
 
94
- Logger.info('[State-Sync] ✅ Success');
95
- })().catch((err) => {
96
- // Re-throw as an unhandled rejection so Node exits with a non-zero
97
- // status. Per orchestration-error-handling rule, orchestrator CLIs MUST
98
- // surface failures via throw rather than Logger.fatal so a stubbed
99
- // process.exit (in tests) does not silently mask the error.
100
- throw new Error(err.message);
101
- });
86
+ // Optional secondary label removal alongside the state transition
87
+ // (e.g. clear `status::blocked` when transitioning back to ready).
88
+ if (removeLabel) {
89
+ await ticketing.updateTicket(ticketId, {
90
+ labels: { remove: [removeLabel] },
91
+ });
92
+ }
93
+
94
+ Logger.info('[State-Sync] ✅ Success');
95
+ return { ticketId, state, removedLabel: removeLabel ?? null };
102
96
  }
97
+
98
+ runAsCli(import.meta.url, () => runUpdateTicketState(), {
99
+ source: 'update-ticket-state',
100
+ usage: USAGE,
101
+ });
@@ -32,7 +32,7 @@ has to be settled here instead.
32
32
  Run each `audit-*` workflow with **no** `--paths` and no change-set filter, so
33
33
  the whole target-set union is audited rather than whatever a recent branch
34
34
  happened to touch. Each lens writes its report to
35
- `temp/audits/audit-<lens>-results.md`.
35
+ `temp/audits/audit-results/audit-<lens>-results.md`.
36
36
 
37
37
  A sweep scoped to a change set is not a sweep: it re-reports the same recent
38
38
  files every cycle and never reaches the code nobody has touched in a year,
@@ -11,7 +11,7 @@ concrete element and the success criterion it violates. Default to **static**
11
11
  detection over source; escalate to a **runtime** pass only when a live target is
12
12
  configured. The shared lens machinery lives in
13
13
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
14
- `{{auditOutputDir}}/audit-accessibility-results.md`. Extra finding fields:
14
+ `{{auditResultsDir}}/audit-accessibility-results.md`. Extra finding fields:
15
15
  **WCAG:** (the violated success criterion, e.g. `1.1.1 Non-text Content (A)`)
16
16
  and **Evidence:** (`measured | static` + the observable; single-run runtime
17
17
  numbers are tagged `provisional`). The report adds a **Runtime Verification**
@@ -18,7 +18,7 @@ That is this lens's central target; structural tidiness is the cheap part.
18
18
 
19
19
  The shared lens machinery lives in
20
20
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
21
- `{{auditOutputDir}}/audit-adrs-results.md`. Dimension values:
21
+ `{{auditResultsDir}}/audit-adrs-results.md`. Dimension values:
22
22
  `Decision Drift | Supersede-Chain Integrity | Structure & Status Hygiene |
23
23
  Missing Decision | Layout Conformance`; the report adds a **Decision Coverage**
24
24
  table.
@@ -8,7 +8,7 @@ You are a Staff Software Engineer & Architecture Reviewer performing a read-only
8
8
  review that prioritizes maintainability and readability without altering
9
9
  external APIs or business logic. The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-architecture-results.md`. Each finding carries an
11
+ `{{auditResultsDir}}/audit-architecture-results.md`. Each finding carries an
12
12
  **Impact:** and a **Category:** (`Quick Win | Structural Change`); the report
13
13
  adds a **Triage Summary** (Quick Wins / Structural Changes) and an
14
14
  **Architecture Guardrail Coverage** section.
@@ -12,7 +12,7 @@ the loop that burns the measured debt **down** and tightens the floors behind
12
12
  it, so a repo can hold a floor it cleared years ago and never notice. This lens
13
13
  is that loop's read-only entry point. The shared lens machinery lives in
14
14
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
15
- `{{auditOutputDir}}/audit-baselines-results.md`. Dimension values:
15
+ `{{auditResultsDir}}/audit-baselines-results.md`. Dimension values:
16
16
  `Dead Instrument | Staleness | Hotspot Cluster | Trend Drift | Tightening Headroom`.
17
17
 
18
18
  > **Value-free titles (mandatory).** A finding title MUST NOT embed a measured
@@ -107,8 +107,13 @@ Two envelope-level reads come **before** any finding:
107
107
  - `baselineExists` is `false`, or `parseError` is non-null — the instrument
108
108
  cannot be read at all.
109
109
  - `deadIgnoreGlobs` is non-empty — a configured ignore pattern matches zero
110
- files. It protects nothing today and silently exempts the next file that
111
- happens to match it.
110
+ files that **its own gate's scanner walk** would visit: that gate's
111
+ `targetDirs`, minus the directories its scanner skips. It protects nothing
112
+ today and silently exempts the next file that happens to match it. A glob
113
+ aimed at a generated or installed tree (`node_modules`, `dist`, `build`,
114
+ `out`, `coverage`, `.next`) that the scanner would visit is never reported
115
+ dead — it is load-bearing after an install or build, even when the tree is
116
+ absent now.
112
117
 
113
118
  Grade a stub or unenforced gate **Medium**, a `parseError` on an enforced
114
119
  gate **High** (delivery reads that file every run), a dead glob **Low**.
@@ -252,7 +257,7 @@ shared finding skeleton, fingerprints each finding for dedupe, and groups the
252
257
  batch:
253
258
 
254
259
  ```bash
255
- node .agents/scripts/audit-to-stories.js --scan --glob temp/audits/audit-baselines-results.md --out temp/audits/audit-to-stories-plan.json
260
+ node .agents/scripts/audit-to-stories.js --scan --glob temp/audits/audit-results/audit-baselines-results.md --out temp/audits/audit-to-stories-plan.json
256
261
  ```
257
262
 
258
263
  Report the plan path and the group count; the converter owns everything
@@ -8,7 +8,7 @@ You are a Principal Software Engineer & Code Quality Lead auditing
8
8
  maintainability — code smells, technical debt, and clean-code violations (SOLID,
9
9
  DRY, KISS) that hinder long-term velocity. The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-clean-code-results.md`. The report adds a **Dead Code
11
+ `{{auditResultsDir}}/audit-clean-code-results.md`. The report adds a **Dead Code
12
12
  Inventory** table (File / Symbol / Type / Estimated LOC) and a **Technical Debt
13
13
  Backlog** section.
14
14
 
@@ -11,7 +11,7 @@ schema silently disagree, where an assumed invariant is not enforced by a
11
11
  constraint, where a migration is unsafe against a live database, and where
12
12
  schema exists that nothing uses. The shared lens machinery lives in
13
13
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
14
- `{{auditOutputDir}}/audit-data-model-results.md`. Dimension values:
14
+ `{{auditResultsDir}}/audit-data-model-results.md`. Dimension values:
15
15
  `Drift | Constraint Completeness | Migration Hygiene | Type Fidelity |
16
16
  Access-Pattern Fit | Dead Schema`. Extra finding field: **Evidence:** (a
17
17
  reference-count or introspection command plus a `measured`, `static`, or
@@ -8,7 +8,7 @@ You are a DevOps Engineer & Security Researcher managing the dependency
8
8
  lifecycle — outdated, vulnerable, or bloated packages and a safe upgrade path.
9
9
  The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-dependencies-results.md`. Dimension values:
11
+ `{{auditResultsDir}}/audit-dependencies-results.md`. Dimension values:
12
12
  `Security Fix | Removal | Engine Drift | Major Upgrade | Supply-chain`. The
13
13
  report adds a **Health Summary**, an **Upgrade Batches** section, and a
14
14
  **Recommended Removals/Replacements** list.
@@ -8,7 +8,7 @@ You are a Principal DevOps Engineer & Infrastructure Architect auditing this
8
8
  repo's DevOps infrastructure, DX tooling, and CI/CD pipelines for
9
9
  inefficiencies, security risks, and modernization opportunities. The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-devops-results.md`. The report adds a **Proposed
11
+ `{{auditResultsDir}}/audit-devops-results.md`. The report adds a **Proposed
12
12
  Implementation Roadmap** section (a logical, phased plan).
13
13
 
14
14
  ## Scope
@@ -12,7 +12,7 @@ mode) catch broken links and generator drift — they cannot tell whether the
12
12
  prose still describes how the code actually behaves. That semantic
13
13
  verification is this lens's job. The shared lens machinery lives in
14
14
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
15
- `{{auditOutputDir}}/audit-documentation-results.md`. Each finding carries a
15
+ `{{auditResultsDir}}/audit-documentation-results.md`. Each finding carries a
16
16
  **Category:** (`Broken Instruction | Stale Description | Missing Coverage |
17
17
  Generator Drift | Link Integrity | History Bloat | Contradiction | Authority
18
18
  Drift`); the report adds a **Target Set Coverage** table.
@@ -11,7 +11,7 @@ advisory, a skipped test, a `KNOWN_*` list. Nothing else checks whether they
11
11
  still earn their place, so they pile up unnoticed. This lens is that check,
12
12
  and it is read-only. The shared lens machinery lives in
13
13
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
14
- `{{auditOutputDir}}/audit-exceptions-results.md`. Dimension values:
14
+ `{{auditResultsDir}}/audit-exceptions-results.md`. Dimension values:
15
15
  `Dead | Expired | Orphaned | Unjustified | Fix Properly`.
16
16
 
17
17
  > **Value-free titles (mandatory).** A finding title MUST NOT embed a count,
@@ -199,7 +199,7 @@ substituting the envelope's own values for the angle-bracketed slots:
199
199
  ## Step 5: Hand off to `/audit-to-stories`
200
200
 
201
201
  ```bash
202
- node .agents/scripts/audit-to-stories.js --scan --glob temp/audits/audit-exceptions-results.md --out temp/audits/audit-to-stories-plan.json
202
+ node .agents/scripts/audit-to-stories.js --scan --glob temp/audits/audit-results/audit-exceptions-results.md --out temp/audits/audit-to-stories-plan.json
203
203
  ```
204
204
 
205
205
  Report the plan path and the group count; the converter owns everything
@@ -10,7 +10,7 @@ a phone and a tablet — and does anything in the suite actually verify that it
10
10
  does? Default to **static** detection over source; escalate to a **runtime**
11
11
  viewport pass only when a live target is configured. The shared lens machinery lives in
12
12
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
13
- `{{auditOutputDir}}/audit-mobile-results.md`. Dimension values:
13
+ `{{auditResultsDir}}/audit-mobile-results.md`. Dimension values:
14
14
  `Layout & Viewport | Touch Ergonomics | Responsive Assets | Mobile Verification`.
15
15
  Extra finding field: **Evidence:** (`measured | static` + the observable;
16
16
  single-run runtime numbers are tagged `provisional`).
@@ -18,7 +18,7 @@ not exist). This is **mechanism, not content**: the lens reads the consumer's
18
18
  configured route tree and nav-registry SSOT and is a **silent no-op when neither
19
19
  is configured**. The shared lens machinery lives in
20
20
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
21
- `{{auditOutputDir}}/audit-navigability-results.md`. Dimension values:
21
+ `{{auditResultsDir}}/audit-navigability-results.md`. Dimension values:
22
22
  `Orphaned Route | Dead Nav Href`; extra finding fields **Route / Door:** and
23
23
  **Persona(s):** (identifiers only — never full route bodies or persona PII).
24
24
 
@@ -13,7 +13,7 @@ only the dimensions that apply, declaring the rest inapplicable); and
13
13
  **interleaving correctness is a performance concern** (races have been the most
14
14
  expensive defects this repo shipped). The shared lens machinery lives in
15
15
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
16
- `{{auditOutputDir}}/audit-performance-results.md`. Extra finding field:
16
+ `{{auditResultsDir}}/audit-performance-results.md`. Extra finding field:
17
17
  **Evidence:** [a Step 0 repro command (or quoted code path) + a `measured` or
18
18
  `estimated` tag]. The report adds a **Low-Hanging Fruit** section.
19
19
 
@@ -8,7 +8,7 @@ You are a Data Privacy Officer & Security Engineer finding accidental logging,
8
8
  insecure storage, or unnecessary collection of PII, and checking GDPR/CCPA
9
9
  compliance. The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-privacy-results.md`. This lens labels the severity
11
+ `{{auditResultsDir}}/audit-privacy-results.md`. This lens labels the severity
12
12
  axis **Impact** and uses dimension values `Leaky Log | Insecure Storage | Data
13
13
  Over-collection`; its report adds a **Privacy Scorecard** section (Data
14
14
  Encryption / Logging Safety / Minimization: Pass/Fail/Partial).
@@ -9,7 +9,7 @@ infrastructure, coverage, flaky tests, mocking strategy, and test-pyramid
9
9
  balance — and, in Story-scoped mode, evaluating the implemented tests against
10
10
  the Story under audit. The shared lens machinery lives in
11
11
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
12
- `{{auditOutputDir}}/audit-quality-results.md`. Each finding carries a
12
+ `{{auditResultsDir}}/audit-quality-results.md`. Each finding carries a
13
13
  **Category:** (`Flakiness | Coverage | Performance | Mocking | Test Plans`); the
14
14
  report adds a **Test Strategy Assessment** table (Unit / Integration / E2E /
15
15
  Test Plans / Property-Based Testing: Healthy / Needs Work / Missing, or `N/A`
@@ -8,7 +8,7 @@ You are a Cybersecurity Architect conducting a comprehensive security review
8
8
  (OWASP Top 10, insecure configs, attack vectors). The shared lens machinery
9
9
  lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-security-results.md`. Extra finding fields: **CWE
11
+ `{{auditResultsDir}}/audit-security-results.md`. Extra finding fields: **CWE
12
12
  ID:**, **Baseline MUST:** (the violated MUST, or `none — trust-boundary
13
13
  class`) and **Boundary:** (Step 2's six-clause gate). The report adds a
14
14
  **Defensive Recommendations** section (3–5 hardening headers/configs/
@@ -9,7 +9,7 @@ You are a Senior Technical SEO & Generative Engine Optimization (GEO) Specialist
9
9
  and content-level improvements that increase discoverability in both traditional
10
10
  search indexes and AI answer engines. The shared lens machinery lives in
11
11
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
12
- `{{auditOutputDir}}/audit-seo-results.md`. Each finding carries a **Category:**
12
+ `{{auditResultsDir}}/audit-seo-results.md`. Each finding carries a **Category:**
13
13
  (`SEO | GEO | Core Web Vitals | Crawlability`); the report adds a **Detailed
14
14
  Audit Table** (an at-a-glance index — not machine-parsed; every row MUST also
15
15
  have a full Detailed Findings entry) and a **GEO-Specific Recommendations**
@@ -9,7 +9,7 @@ a production release candidate — rollback & recovery paths, observability &
9
9
  instrumentation, resilience & failure handling, and runbooks & operational
10
10
  docs. The shared lens machinery lives in
11
11
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
12
- `{{auditOutputDir}}/audit-sre-results.md`. Each finding carries a **Category:**
12
+ `{{auditResultsDir}}/audit-sre-results.md`. Each finding carries a **Category:**
13
13
  (`Rollback & Recovery | Observability | Resilience | Runbooks`); the report adds
14
14
  a **Release Readiness Checklist** table (per-category ✅ Clear / ⚠️ Issues
15
15
  Found).
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  description:
3
3
  Convert findings produced by the audit-* workflows into actionable
4
- GitHub Stories. Reads temp/audits/audit-*-results.md, groups findings
4
+ GitHub Stories. Reads temp/audits/audit-results/audit-*-results.md, groups findings
5
5
  cross-audit, deduplicates against existing Issues by fingerprint, and
6
6
  either chains into /mandrel-plan --seed-file or opens standalone Stories.
7
7
  ---
@@ -15,7 +15,9 @@ Engineering Lead
15
15
  ## Context
16
16
 
17
17
  The `audit-*` workflows each produce a structured `audit-<dimension>-results.md`
18
- report under `temp/audits/`. Every `### Finding` block in those reports
18
+ report under `temp/audits/audit-results/` — a folder that holds lens reports
19
+ and nothing else; this workflow's own plan, seed and stories files stay one
20
+ level up in `temp/audits/`. Every `### Finding` block in those reports
19
21
  already carries the fields a Story body needs (Severity / Impact,
20
22
  Dimension / Category, Location, Current State, Recommendation, Acceptance
21
23
  signal).
@@ -26,7 +28,7 @@ They remain read-only emitters of audit reports.
26
28
  ## Prerequisites
27
29
 
28
30
  1. At least one `audit-*-results.md` file under
29
- `temp/audits/` (or the path passed as the argument). Run a
31
+ `temp/audits/audit-results/` (or the path passed as the argument). Run a
30
32
  `/audit-<dimension>` first if none are present.
31
33
  2. `GITHUB_TOKEN` or `gh auth status` clean — the dedupe and create
32
34
  steps both call GitHub.
@@ -38,7 +40,7 @@ They remain read-only emitters of audit reports.
38
40
 
39
41
  `/audit-to-stories [audit-file-or-glob]`
40
42
 
41
- - No argument → scans `temp/audits/audit-*-results.md`.
43
+ - No argument → scans `temp/audits/audit-results/audit-*-results.md`.
42
44
  - Single file path or glob → restricts the scan to that input.
43
45
 
44
46
  ## Phase 1 — Discover & parse
@@ -330,7 +332,7 @@ result as a JSON array, and pass it:
330
332
 
331
333
  ```bash
332
334
  node .agents/scripts/audit-to-stories.js --scan --no-provider \
333
- --issues-file temp/audits/issues.json --glob "temp/audits/audit-*-results.md"
335
+ --issues-file temp/audits/issues.json --glob "temp/audits/audit-results/audit-*-results.md"
334
336
  ```
335
337
 
336
338
  Real `skip-open` / `skip-reoccurring` classifications come back. The corpus is
@@ -403,12 +405,12 @@ that delivers the whole sweep (`/mandrel-deliver <epicId>`).
403
405
  To run an unattended maintenance sweep, `/schedule` a nightly (or weekly)
404
406
  job that (1) runs the relevant `audit-*` lens workflows full-scope — no
405
407
  `--paths`, no change-set filter, so the whole target-set union is audited —
406
- writing their `temp/audits/audit-*-results.md` reports, then (2) invokes the
408
+ writing their `temp/audits/audit-results/audit-*-results.md` reports, then (2) invokes the
407
409
  CLI's **`--auto` mode** over those results:
408
410
 
409
411
  ```bash
410
412
  node .agents/scripts/audit-to-stories.js --auto [--dry-run] \
411
- [--glob "temp/audits/audit-*-results.md"] [--severity <floor>]
413
+ [--glob "temp/audits/audit-results/audit-*-results.md"] [--severity <floor>]
412
414
  ```
413
415
 
414
416
  The routine shape is **lenses full-scope → dry-run → live**:
@@ -8,7 +8,7 @@ You are a Lead Product Designer & Frontend Architect evaluating the frontend for
8
8
  UI consistency, UX best practices, and adherence to the project's design system,
9
9
  ensuring the app feels premium and cohesive. The shared lens machinery lives in
10
10
  [`helpers/audit-lens-core.md`](helpers/audit-lens-core.md). Write the report to
11
- `{{auditOutputDir}}/audit-ux-ui-results.md`. Dimension values:
11
+ `{{auditResultsDir}}/audit-ux-ui-results.md`. Dimension values:
12
12
  `Visual Consistency | UX Best Practices | Accessibility`; the report adds a
13
13
  **Micro-animation Opportunities** section.
14
14
 
@@ -87,12 +87,15 @@ one — and name the detected set in the Executive Summary. Then:
87
87
  ## Report envelope & finding-block skeleton {#report-envelope}
88
88
 
89
89
  Each lens writes exactly one structured Markdown report to
90
- `{{auditOutputDir}}/audit-<lens>-results.md` (the lens preamble names its own
90
+ `{{auditResultsDir}}/audit-<lens>-results.md` (the lens preamble names its own
91
91
  path). A dispatched run gets the anchor substituted; on a manual
92
- `/audit-<lens>` invocation it renders the literal string `{{auditOutputDir}}`.
93
- Treat that literal as `<project.paths.tempRoot>/audits` — **`temp/audits` by
94
- default** — and write the report there, mirroring the `{{changedFiles}}`
95
- manual-run rule above. The report MUST include every section its lens
92
+ `/audit-<lens>` invocation it renders the literal string `{{auditResultsDir}}`.
93
+ Treat that literal as `<project.paths.tempRoot>/audits/audit-results` —
94
+ **`temp/audits/audit-results` by default**, created if absent — and write the
95
+ report there, mirroring the `{{changedFiles}}` manual-run rule above. That
96
+ folder holds lens reports only: any other artifact a lens writes (an
97
+ introspected schema, `perf-baseline.json`, profiler output) goes under
98
+ `{{auditOutputDir}}`, read on a manual run as `<project.paths.tempRoot>/audits`. The report MUST include every section its lens
96
99
  template mandates — write `_No findings._` rather than omitting a section — and
97
100
  always an
98
101
  `## Executive Summary` and a `## Detailed Findings` section. The Executive
@@ -12,8 +12,7 @@ description: >-
12
12
  > covers the happy path. Situational material (lease preflight, recovery
13
13
  > routers, merge-wait budgets, CI remediation) stays on demand in
14
14
  > [`deliver-story-reference.md`](deliver-story-reference.md) and
15
- > [`deliver-reference.md`](deliver-reference.md); read those **only** when an
16
- > envelope or a failure routes you there.
15
+ > [`deliver-reference.md`](deliver-reference.md); read those only when routed.
17
16
 
18
17
  ## 1. Dispatch — where the engine runs
19
18
 
@@ -25,8 +24,8 @@ rule produces it:
25
24
  sibling, and a one-Story run has none.
26
25
  2. **Every other run is `subagent`**, however trivial each Story's shape.
27
26
 
28
- `inline` removes the `story-worker` boot and nothing else — not the verdict
29
- owner, which the profile alone names (§ 3). **`subagent` and `inline` run the same engine**: same gates,
27
+ `inline` drops only the `story-worker` boot — not the verdict owner, which
28
+ the profile alone names (§ 3). **`subagent` and `inline` run the same engine**: same gates,
30
29
  same PR to `main`, same terminal envelope.
31
30
 
32
31
  ## 2. Engine invariants
@@ -154,7 +153,7 @@ respawned.
154
153
  | `status` | Exit | Meaning | You do |
155
154
  | --- | --- | --- | --- |
156
155
  | `landed` | 0 | PR merged, `agent::done`, tail ran (`tail.*: false` degrades the report, not the land) | Relay it. Done. |
157
- | `pending` | 3 | **Resumable, not a failure** — the bounded wait expired healthy, or a human owns the merge. Nothing was mutated. | Run `nextCommand`. |
156
+ | `pending` | 3 | **Resumable, not a failure** — the bounded wait expired healthy, or a human owns the merge. Nothing was mutated. | Run `nextCommand`. Each beat: run every `confirm[]` verbatim, pre-epilogue. |
158
157
  | `blocked` | 1 | Hard block; `blocked.blockClass` names it | `checks-failed` → fix + resume; else relay |
159
158
  | `failed` | 1 | A phase crashed; `phase` names which | Diagnose, fix, re-run close. `base-sync` conflict (files in `failure.reason`): resolve in `workCwd`, commit, run `nextCommand`; block only if you cannot |
160
159
 
@@ -228,8 +228,8 @@ separate (`delivery.mergeWatch.*`):
228
228
  that precede it. Expiry → `pending`. Pass `--max-wait-seconds <n>` to raise
229
229
  it when your host has no such ceiling and you want to land in one block.
230
230
  - **`maxBudgetSeconds`** (default 3600) bounds the **cumulative** wait across
231
- resumes, anchored at the PR's `createdAt` so resuming does not restart the
232
- clock. Exhausting _this_ is the genuine give-up → `blocked`.
231
+ resumes, anchored at the latest auto-merge arm (else `createdAt`): a resume
232
+ keeps the clock, a re-arming re-close starts fresh. Exhausting _this_ is the genuine give-up → `blocked`.
233
233
 
234
234
  The wait probes the checks every poll: a red required check fails fast as
235
235
  `checks-failed` instead of burning the budget, and a PR that falls behind its
@@ -623,8 +623,8 @@ tool-invocation ceiling:
623
623
 
624
624
  - It is **resumable**: no label was mutated, no `merge.unlanded` was emitted,
625
625
  and `nextCommand` names the one command that continues it. The cumulative
626
- budget is anchored at the PR's `createdAt`, so resuming does not restart the
627
- clock.
626
+ budget is anchored at the latest auto-merge arm, which a resume does not
627
+ move, so resuming does not restart the clock.
628
628
  - It is **not** a park. Returning `pending` because you would rather not wait
629
629
  is the no-park failure mode wearing a schema. Return it only
630
630
  when the bound genuinely expired, or a human owns the merge.
@@ -102,7 +102,9 @@ to an attended run.
102
102
  `effort` is non-null, pass it as that Agent call's `effort`; when it is
103
103
  `null`, omit the parameter so the worker inherits the session. Run each
104
104
  `close[]` entry's command foreground and serialized as hand-offs arrive.
105
- An empty
105
+ Run every `confirm[]` command verbatim on each beat, before the epilogue:
106
+ a closed Story still owing its `agent::done` flip and land tail. It never
107
+ gates `done`. An empty
106
108
  `ready` with work in flight means "waiting", so beat again — **unless
107
109
  `stalledDispatch[]` is non-empty**: those ids were dispatched but still
108
110
  read `agent::ready`, which is a live init window the first beats or two
package/README.md CHANGED
@@ -30,10 +30,13 @@ directory, where your assistant reads it.
30
30
  | `scripts/` | Deterministic Node.js CLIs that do the mechanical work — GitHub state changes, worktrees, quality gates — so the model judges and the scripts execute. |
31
31
  | `schemas/`, `templates/` | JSON Schema contracts and the prompt/planning templates the workflows fill in. |
32
32
  | `docs/` | Consumer reference: [configuration](.agents/docs/configuration.md) (every `.agentrc.json` key), the [workflow index](.agents/docs/workflows.md), and operator runbooks. |
33
+ | `audit-checklists/` | Generated per-lens authoring checklists, threaded into a Story's write-time checklist when its change set matches an audit lens. |
34
+ | `mods/` | Beta, opt-in Claude Code mods — see the [mods runbook](.agents/docs/runbooks/mods.md). |
33
35
  | `local/` | Yours. `mandrel sync` never touches it, so project-specific skills and workflow fragments survive upgrades. |
34
36
 
35
- Project settings live in `.agentrc.json` at your repo root. The full tour of
36
- the bundle is in [`.agents/README.md`](.agents/README.md).
37
+ Project settings live in `.agentrc.json` at your repo root. The table above is
38
+ a partial list (it omits single files such as `starter-agentrc.json`); the full
39
+ tour of the bundle is in [`.agents/README.md`](.agents/README.md).
37
40
 
38
41
  ## The core loop: plan, then deliver
39
42
 
@@ -82,7 +85,7 @@ The full process — labels, state, review, troubleshooting — is in
82
85
 
83
86
  The `/audit-*` workflows are read-only lenses that examine the codebase along
84
87
  one dimension and write a findings report (severity, location, recommendation,
85
- acceptance signal) under `temp/audits/`. They never change code — fixing is a
88
+ acceptance signal) under `temp/audits/audit-results/`. They never change code — fixing is a
86
89
  separate, planned pass.
87
90
 
88
91
  | Area | Lenses |