mandrel 2.120.0 → 2.122.0

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 (41) hide show
  1. package/.agents/agents/acceptance-critic.md +5 -1
  2. package/.agents/agents/story-worker.md +1 -1
  3. package/.agents/docs/agentrc-reference.json +1 -0
  4. package/.agents/docs/configuration.md +1 -0
  5. package/.agents/instructions.md +3 -3
  6. package/.agents/schemas/acceptance-eval-verdict.schema.json +6 -1
  7. package/.agents/schemas/agentrc.schema.json +9 -0
  8. package/.agents/scripts/acceptance-eval.js +125 -13
  9. package/.agents/scripts/audit-to-stories.js +21 -0
  10. package/.agents/scripts/evidence-gate.js +15 -2
  11. package/.agents/scripts/lib/audit-to-stories/audit-label-taxonomy.js +7 -3
  12. package/.agents/scripts/lib/audit-to-stories/audit-lenses.js +18 -1
  13. package/.agents/scripts/lib/baselines/crap-preview-scan.js +1 -0
  14. package/.agents/scripts/lib/baselines/git-base.js +43 -0
  15. package/.agents/scripts/lib/baselines/kinds/_crap-new-method-gate.js +29 -5
  16. package/.agents/scripts/lib/baselines/kinds/crap.js +7 -1
  17. package/.agents/scripts/lib/baselines/kinds/kind-factory.js +81 -52
  18. package/.agents/scripts/lib/close-validation/projections/advisories.js +1 -0
  19. package/.agents/scripts/lib/close-validation/projections/crap.js +3 -0
  20. package/.agents/scripts/lib/config/gates/crap.schema.js +7 -0
  21. package/.agents/scripts/lib/config/quality.js +13 -6
  22. package/.agents/scripts/lib/generated/agentrc-validator.js +2 -2
  23. package/.agents/scripts/lib/orchestrate/classify.js +30 -14
  24. package/.agents/scripts/lib/orchestrate/inventory.js +10 -6
  25. package/.agents/scripts/lib/orchestration/acceptance-eval-decision.js +65 -6
  26. package/.agents/scripts/lib/orchestration/check-baselines/phases/compare.js +34 -2
  27. package/.agents/scripts/lib/orchestration/check-baselines/phases/fail-detail.js +7 -2
  28. package/.agents/scripts/lib/orchestration/handoff-gates.js +2 -0
  29. package/.agents/scripts/lib/orchestration/preflight-lanes.js +2 -0
  30. package/.agents/scripts/lib/orchestration/ticket-validator.js +35 -0
  31. package/.agents/scripts/lib/orchestration/verify-credit.js +144 -0
  32. package/.agents/scripts/lib/validation-evidence.js +27 -0
  33. package/.agents/workflows/audit-to-stories.md +6 -3
  34. package/.agents/workflows/helpers/acceptance-self-eval.md +9 -1
  35. package/.agents/workflows/helpers/deliver-digest.md +7 -8
  36. package/.agents/workflows/helpers/deliver-story-reference.md +1 -1
  37. package/.agents/workflows/helpers/plan-reference.md +3 -1
  38. package/.agents/workflows/mandrel-orchestrate.md +4 -0
  39. package/.agents/workflows/memory-consolidate.md +1 -1
  40. package/docs/CHANGELOG.md +27 -0
  41. package/package.json +1 -1
@@ -76,18 +76,17 @@ skeleton — it derives § 3's `files` and `verdictOwner` with the same function
76
76
  `node <main-repo>/.agents/scripts/acceptance-eval.js --story <storyId> --init
77
77
  --cwd <workCwd>`
78
78
 
79
- It writes one empty record per `acceptance[]` item and prints its
80
- path. The owner fills **every** record,
81
- scored against that change set with `verify[]` output as evidence, and scores
82
- it in **one** gate call. Bounded by `delivery.acceptanceEval.maxRounds`
79
+ It writes one empty record per `acceptance[]` item and prints its path. The
80
+ owner fills **every** record against that change set, `verify[]` output as
81
+ evidence, and scores it in **one** gate call. Bounded by `delivery.acceptanceEval.maxRounds`
83
82
  (default 2; `0` scores once with no redraft).
84
83
 
85
84
  `node <main-repo>/.agents/scripts/acceptance-eval.js --story <storyId>
86
- --verdict <verdict-path>`
85
+ --verdict <verdict-path> --cwd <workCwd>`
87
86
 
88
- The gate reads the Story's `acceptance[]` count itself and rejects, **before**
89
- scoring and consuming no round, a verdict whose `criteria[]` length differs
90
- or with an unfilled record (naming its indices).
87
+ Refused before scoring (no round): a `criteria[]` length off the
88
+ `acceptance[]` count, an unfilled record, an uncited `verify[]` entry. The
89
+ gate runs scoped entries in `--cwd`.
91
90
 
92
91
  `proceed` → § 5. `redraft` → one more round inside the cap. `block` → **do
93
92
  not close**: post a `friction` comment and flip `agent::blocked`.
@@ -184,7 +184,7 @@ criteria, then transition the Story to `agent::blocked`:
184
184
 
185
185
  ```bash
186
186
  node .agents/scripts/diagnose-friction.js --story <storyId> \
187
- --cmd node .agents/scripts/acceptance-eval.js --story <storyId> --verdict <verdict-path>
187
+ --cmd node .agents/scripts/acceptance-eval.js --story <storyId> --verdict <verdict-path> --cwd <workCwd>
188
188
  node .agents/scripts/update-ticket-state.js --ticket <storyId> --state agent::blocked
189
189
  ```
190
190
 
@@ -381,7 +381,9 @@ half with the second suppressed. Either way every gate runs before the first
381
381
  where the second half is read:
382
382
 
383
383
  **Hard — the run refuses:** a body that does not parse, a ticket that is not
384
- a Story, an empty `acceptance[]`, an unknown or cyclic `depends_on`, the
384
+ a Story, an empty `acceptance[]`, a `verify[]` entry that can exit 0 after a
385
+ failure (`|| true`, `|| :`, `|| break`, `|| continue`, or a loop with no
386
+ failing `exit`; use `|| exit 1`), an unknown or cyclic `depends_on`, the
385
387
  same-wave collision refusal at N>1, a supersede claim on a non-source id or
386
388
  one claimed twice, and a `deletes` entry naming a path absent at base.
387
389
  Commit subjects are not scanned — the `commit-msg` hook and
@@ -97,6 +97,10 @@ annotation, which the inventory already carries.
97
97
  - **`green-not-queued`**: arm auto-merge only when `arm-auto-merge` is
98
98
  granted and no release veto holds (`veto` is not `held`); otherwise raise
99
99
  once.
100
+ - **`checks-cancelled`**: a required check was cancelled, went stale or
101
+ awaits action, so it never passed. Never arm auto-merge or requeue it, and
102
+ never message the owner about a failure; raise it once naming the halted
103
+ checks, so the operator decides whether to re-run them.
100
104
  - **`waiting-on-runner`**: no action. Report the `runners` backlog, and hold
101
105
  back optional load while it is saturated.
102
106
  - **`stale`**: raise once.
@@ -103,7 +103,7 @@ the pass; nothing in the framework reads it back.
103
103
  The `/mandrel-plan` Phase 0 advisory re-arms on exactly one condition:
104
104
  `MEMORY.md` exceeding the fixed 24576-byte index ceiling —
105
105
  the harness cap past which the index it loads is truncated. Pool size and
106
- stamp age never trigger it (Story #5312 retired those arms): a pass that
106
+ stamp age never trigger it: a pass that
107
107
  rewrites long index lines short quiets it without pruning an entry.
108
108
 
109
109
  Write it **only** after Gate #2 — the stamp asserts an operator reviewed the
package/docs/CHANGELOG.md CHANGED
@@ -15,6 +15,33 @@ All notable changes to this project will be documented in this file.
15
15
  -->
16
16
  <!-- markdownlint-disable-file MD004 MD012 MD037 -->
17
17
 
18
+ ## [2.122.0](https://github.com/dsj1984/mandrel/compare/mandrel-v2.121.0...mandrel-v2.122.0) (2026-11-13)
19
+
20
+
21
+ ### Added
22
+
23
+ * audit-to-stories: a non-canonical report's findings carry audit::custom so indexed dedup sees their Stories ([#6081](https://github.com/dsj1984/mandrel/issues/6081)) ([#6082](https://github.com/dsj1984/mandrel/issues/6082)) ([896a468](https://github.com/dsj1984/mandrel/commit/896a4687157ddd5fe4eff000b7b440c2421ed6fe))
24
+
25
+
26
+ ### Fixed
27
+
28
+ * orchestrate-inventory stops reporting cancelled or timed-out required checks as green ([#6084](https://github.com/dsj1984/mandrel/issues/6084)) ([#6085](https://github.com/dsj1984/mandrel/issues/6085)) ([2f1e16b](https://github.com/dsj1984/mandrel/commit/2f1e16b73cd9433ecb18a53bee43fbf4e53ebb2c))
29
+
30
+ ## [2.121.0](https://github.com/dsj1984/mandrel/compare/mandrel-v2.120.0...mandrel-v2.121.0) (2026-11-12)
31
+
32
+
33
+ ### Added
34
+
35
+ * new and renamed files are judged by the coverage gate, and untested new files get the full CRAP check ([#6068](https://github.com/dsj1984/mandrel/issues/6068)) ([#6076](https://github.com/dsj1984/mandrel/issues/6076)) ([b28ce16](https://github.com/dsj1984/mandrel/commit/b28ce16a6d739f94782df6748632cf3952aae989))
36
+ * plan-persist refuses verify entries that discard their exit status ([#6071](https://github.com/dsj1984/mandrel/issues/6071)) ([#6073](https://github.com/dsj1984/mandrel/issues/6073)) ([9fb3ace](https://github.com/dsj1984/mandrel/commit/9fb3ace0e1ca567034050e6485a57d837742e4a9))
37
+ * the acceptance gate runs the Story's scoped verify[] entries itself and refuses a met verdict without passing evidence ([#6067](https://github.com/dsj1984/mandrel/issues/6067)) ([#6077](https://github.com/dsj1984/mandrel/issues/6077)) ([fbe68cd](https://github.com/dsj1984/mandrel/commit/fbe68cd21a7646db46e3efc72200a8c8a3fc4d95))
38
+
39
+
40
+ ### Fixed
41
+
42
+ * evidence-gate neither spends nor records credit when the working tree differs from HEAD ([#6070](https://github.com/dsj1984/mandrel/issues/6070)) ([#6074](https://github.com/dsj1984/mandrel/issues/6074)) ([c0bf74e](https://github.com/dsj1984/mandrel/commit/c0bf74e78e517e859570c37ad9b88e6dfd7ab1e6))
43
+ * handoff's credit re-recorders skip a dirty tree, sharing evidence-gate's probe ([#6078](https://github.com/dsj1984/mandrel/issues/6078)) ([#6079](https://github.com/dsj1984/mandrel/issues/6079)) ([012f305](https://github.com/dsj1984/mandrel/commit/012f305a0a62c1e569f475f7ef17ead9bf8ab08c))
44
+
18
45
  ## [2.120.0](https://github.com/dsj1984/mandrel/compare/mandrel-v2.119.0...mandrel-v2.120.0) (2026-11-12)
19
46
 
20
47
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "mandrel",
3
- "version": "2.120.0",
3
+ "version": "2.122.0",
4
4
  "description": "Claude Code-first opinionated workflow framework: instructions, skills, rules, and SDLC workflows that govern AI coding assistants.",
5
5
  "files": [
6
6
  ".agents/",