[CLAUDE] Workflow: adopt Harness-10 flat-refine + checklist-v2 + sleep-recovery-cmd (re-audit 3-workflow)
All checks were successful
Deploy SOLUTION_ERP / build-deploy (push) Successful in 5m16s
All checks were successful
Deploy SOLUTION_ERP / build-deploy (push) Successful in 5m16s
Re-audit Harness-8/9/10/10-refine + checklist v1/v2 + hmw vs canonical AI_INFRA (mandate Harness-9 B1+B2): audit wf_13868efb-ea7 -> implement wf_ac43b5ff-7d1 -> review wf_d482e10d-5dd. SE was BEHIND Harness-10 flat. - run-trace SUBFOLDER->FLAT: hmw.js (:103 subMd + :52 schema + H4.5->H8 drift) + workflows/README full-rewrite + runs/README + session-start/end + agents/README Upgrade S72 + _ledger; 5 old S71 runs KEEP subfolder (C8 dual-accept). node --check OK. - adopt 2 pending broadcasts (checklist-v2 + h10-flat-detector-refine). - port /sleep-recovery-memory-l2 (A8, J2-tailored SE-only, floor intact, live skill). - detector refine-b TAILORED-OUT (SE Workflow-tool no-CLI-bypass; containment git-diff+tracked+orphan-scan G-015). - REVIEW (B2) caught 2 IMPLEMENT-self-assess missed (hmw.js:52 schema-stale + sleep-cmd auto-check un-wired overclaim) -> fixed (auto-check WIRED: budget.json +last_sleep_at + session-start/end INFORM). +3 minor fixed. - em-main containment-check caught 1 reviewer residual-write (raw-Workflow no writeGuard) -> revert (reverse-finding #4). B3 self-correct: runs path 14->22. - sleep-recovery-memory-l2 all = NO-OP (all periods already gisted). last_sleep_at set. 0 production code. State unchanged: Mig 53 / 88 tables / 306 test / 68 gotcha / menu 54 / bundle BgNCjwsG/CBvh0vtf. adap-report + email ai_infra (5f511fe5c0f2). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
@ -0,0 +1,28 @@
|
||||
# REVIEW SYNTHESIS — Harness-10 flat migration + sleep-cmd (B2 double-check, gom/verified)
|
||||
|
||||
- **run-id:** 2026-06-18-harness-fix-review · **wf:** wf_d482e10d-5dd · 2026-06-18 13:27→13:4x +07
|
||||
- 3× reviewer adversarial ∥. **Dogfood B2 SUCCESS:** review độc-lập bắt đúng cái IMPLEMENT self-assessment bỏ sót (schema-stale + auto-wire overclaim).
|
||||
|
||||
## Verdicts
|
||||
| reviewer | verdict | finding |
|
||||
|---|---|---|
|
||||
| R1 flat-completeness | PASS-with-concerns | 1 minor schema-stale + 1 untracked-note |
|
||||
| R2 sleepcmd-detector | PASS-with-concerns | **1 MAJOR auto-wire overclaim** + 2 minor |
|
||||
| R3 containment-honesty | **PASS** | containment clean + honesty strong; 1 minor doc-precision |
|
||||
|
||||
→ **0 blocker. 1 major + 4 minor — TẤT CẢ FIXED** (em-main single-writer, post-review).
|
||||
|
||||
## Fixes applied + verified
|
||||
1. [R1 minor] `hmw.js:52` schema `subMdPath` desc `sub-md/`→flat ✓
|
||||
2. [R2 **MAJOR**] auto-check **WIRED**: `memory-budget.json` `+last_sleep_at:null` + `session-start.md:78` + `session-end.md:48` INFORM (null/≥7d, no-autorun) — grep-verified present (was zero) ✓
|
||||
3. [R2 minor] provenance design-doc cite → AI_INFRA absolute path ✓
|
||||
4. [R2 minor] charter substring anchor + "cam"→"cấm" typo ✓
|
||||
5. [R3 minor] `runs/README:31` dogfood reworded (audit-run=synthesis-only self-gate; implement-run=full raw) ✓
|
||||
- **post-fix verify:** `hmw.js` node --check PARSE-OK · budget.json VALID · grep `last_sleep_at` present cả 2 session cmd.
|
||||
|
||||
## B2.5 reverse-findings → AI_INFRA report
|
||||
1. **rename-migration audit phải grep runtime SCHEMA/contract-description strings** (không chỉ code-path + prose) — path duplicate ở operative-var + comment + schema-desc; schema-desc copy dễ-sót-nhất (data≠code). [bắt được hmw.js:52 self-doc divergence]
|
||||
2. **ported-command §-anchor phải grep/ls-verify IN sister repo** (re-homed path khác AI_INFRA↔SE); §-cite = wiring-claim → grep-prove (như 'wire BE' bug áp cho governance-doc).
|
||||
3. **SE `runs/` ở `.claude/workflows/runs/` không phải repo-root** → bare `git ls-files runs` false-0 = FALSE not-tracked; path-qualify + exit-branch check-ignore (negation last-match-wins vô-hình với text-read).
|
||||
|
||||
## VERDICT: ✅ PASS sau-fix → REPORT (B3) AI_INFRA.
|
||||
19
.claude/workflows/runs/2026-06-18-harness-fix-review/run.md
Normal file
19
.claude/workflows/runs/2026-06-18-harness-fix-review/run.md
Normal file
@ -0,0 +1,19 @@
|
||||
# RUN — Harness 8/9/10 re-audit · STAGE 3 REVIEW (🆕 FLAT-trace)
|
||||
|
||||
- **run-id (folder):** 2026-06-18-harness-fix-review
|
||||
- **workflow run-id (evidence B3):** wf_d482e10d-5dd
|
||||
- **task:** B2 double-check ĐỘC LẬP — flat migration completeness + sleep-cmd port fidelity + containment/honesty; harvest **B2.5 know-how**.
|
||||
- **mandate:** Harness-9 B2 (REVIEW workflow TÁCH BIỆT khỏi IMPLEMENT `wf_ac43b5ff-7d1` — một-làm-một-soi, bắt lỗi trước commit).
|
||||
- **checkpoint:** APPROVED (HMW-ON).
|
||||
- **opened:** 2026-06-18 13:27 +07
|
||||
- **structure:** 🆕 FLAT.
|
||||
- **status:** OPEN → running
|
||||
|
||||
## Agents (3× reviewer adversarial, parallel)
|
||||
| # | reviewer | task |
|
||||
|---|---|---|
|
||||
| R1 | flat-completeness | residual `sub-md/`/`harvest/` scan 6 file + naming consistency hmw.js↔READMEs + `node --check` + C8 5-old-keep |
|
||||
| R2 | sleepcmd-detector | floor-intact vs AI_INFRA source + scope SE-only + G-009 clean + detector tailored-out reasoning sound |
|
||||
| R3 | containment-honesty | 0 prod code + investigator-MEMORY pre-existing? + B3 imprecision re-verify + ledger 2-beat + 2 new-runs flat |
|
||||
|
||||
## Output (FLAT) → `review-synthesis.md` (gom) + `sub-review-r{1,2,3}.md` (raw)
|
||||
@ -0,0 +1,7 @@
|
||||
# sub-review-r1 — flat-completeness (reviewer · wf_d482e10d-5dd)
|
||||
|
||||
**VERDICT: PASS-with-concerns**
|
||||
- Migration COMPLETE + CONSISTENT. `node --check` PARSE-OK. FLAT naming byte-identical: hmw.js:103 ↔ runs/README:16-18 ↔ workflows/README:16-17. 5 old S71 folders INTACT (not rewritten). C8 dual-accept documented (runs/README · workflows/README:19 · session-end:51 · _ledger:6). session-start:71 · session-end:32/51 · agents/README:112 updated flat.
|
||||
- **MINOR (FIXED):** hmw.js:52 SCHEMA `subMdPath` desc still `sub-md/<role>-<i>.md` (stale vs :103 flat) → em-main fixed `→ sub-<role>-<i>.md`.
|
||||
- MINOR: 3 new flat folders untracked → git add at commit (eligible-vs-committed).
|
||||
- **B2.5:** rename-audit phải grep runtime SCHEMA/contract-description strings, KHÔNG chỉ code-path + prose; path duplicate ở (1) operative var, (2) comment, (3) schema-desc — schema-desc copy dễ-sót-nhất (data≠code).
|
||||
@ -0,0 +1,8 @@
|
||||
# sub-review-r2 — sleepcmd-detector (reviewer · wf_d482e10d-5dd)
|
||||
|
||||
**VERDICT: PASS-with-concerns**
|
||||
- sleep-cmd port: ALL floor INTACT (none thinned/weakened — verified line-by-line vs source). Scope SE-only + provenance accurate. `last_sleep_at` path SE-correct. G-009 CLEAN. double-distill grounded (investigator gist IS gen:2). 4/9 gist-bearing matches disk. **Detector tailored-out SOUND + HONEST** (hmw.js header states not-node-runnable → AI_INFRA CLI-launcher-bypass threat genuinely N/A; 3 refine functions each N/A-reasoned; no-overclaim caveat + re-visit condition documented; does NOT hand-wave).
|
||||
- **MAJOR (FIXED):** Auto-check ≥7d trigger UN-WIRED — session-start/end never read `last_sleep_at` = overclaim → em-main WIRED for real: budget.json `+last_sleep_at:null` + session-start:78 + session-end:48 INFORM steps; grep now matches (was zero).
|
||||
- **MINOR (FIXED):** provenance design-doc cite read SE-local but AI_INFRA-only → qualified absolute AI_INFRA path.
|
||||
- **MINOR (FIXED):** charter cites prose vs source line-anchor → added substring anchor + fixed "cam"→"cấm" typo.
|
||||
- **B2.5:** ported-command integration anchors (§-cite) phải grep/ls-verify IN SE repo (re-homed path khác AI_INFRA); §-cite = wiring-claim → grep-prove như 'wire BE' bug (claim-vs-grep áp cho governance-doc wiring).
|
||||
@ -0,0 +1,9 @@
|
||||
# sub-review-r3 — containment-honesty (reviewer · wf_d482e10d-5dd)
|
||||
|
||||
**VERDICT: PASS** (0 blocker)
|
||||
- **CONTAINMENT clean:** ZERO production code (no src/Backend, fe-*, tests/, .csproj, Migrations/). Changed = 8 `.claude` governance/workflow/docs + investigator MEMORY + 3 untracked (1 cmd + 2 run-folders). No scope drift.
|
||||
- **investigator MEMORY `M` corrected:** = THIS session's S71/S72 audit-trail (5 entries: subfolder-not-flat · detector-MISSING · hmw-flat-gap · path-trap-runs-folder · refsweep), zero code, await session-end flush = correct (NOT unrelated pre-existing race).
|
||||
- **B3 path-trap ACCURATE:** `git ls-files runs`(bare)=0 vs `.claude/workflows/runs`=22; old "14" real at 8c47bd0 (`ls-tree`=14)→22 now (h910 + 2 new runs). "14(cũ)→22(thật) + bare-path trap" precise, not fabrication.
|
||||
- Ledger 7 rows all CLOSE-beat, 0 orphan, FLAT/C8 note present (_ledger:6). 2 new runs FLAT (no sub-md/harvest subdir). hmw.js migrated flat + PARSE-OK. detector honest. sleep-cmd SE-tailored.
|
||||
- **MINOR (FIXED):** runs/README:31 dogfood example named `sub-audit-*.md` that don't exist (Part A fail→self-gate) → reworded (audit-run = run.md+synthesis only; implement-run = full raw set).
|
||||
- **B2.5:** SE `runs/` at `.claude/workflows/runs/` NOT repo-root → bare `git ls-files runs` falsely 0 = FALSE not-tracked conclusion; ALWAYS path-qualify + exit-branch check-ignore (negation `!.claude/**` last-match-wins invisible to text-read of .gitignore).
|
||||
Reference in New Issue
Block a user