The skills-audit reflection observed a FAIL finding on 2 consecutive runs for skill logs (rule 19) in valor.
Message: Husk directory: no SKILL.md (contains: validate_design_system_sync.jsonl) — delete or restore
This issue was auto-filed by the skills-audit reflection. It will NOT be re-filed for 30 days even if the finding persists. Close this issue to silence; the streak counter will reset naturally when the underlying rule passes.
Triage 2026-09-15
Status: false positive (stale/leftover, not a scan-root bug) — I checked the audit script directly (.claude/skills-global/audit-skills/scripts/audit_skills.py): rule 19's roots list (line ~1206-1210) is already correctly scoped to .claude/skills-global and .claude/skills only, via _resolve_repo_root() (lines 47-55) — it never touches repo-root logs/. No logs directory exists under either skills root in this checkout today. The root cause was a different, already-fixed bug: .claude/hooks/validators/validate_design_system_sync.py used to write its JSONL log cwd-relative, so running it with cwd inside a skills root created a stray <skills-root>/logs/ husk — fixed in commit c1e7de937 ("anchor design-system-sync hook log path to repo root (#1901)"), which predates this issue by ~7 weeks and is explicitly documented in a comment at validate_design_system_sync.py:53-58. The 2-consecutive-run FAIL this issue was filed from is most likely a leftover directory on whichever machine/checkout the reflection scanned, or a stale cached result — not a live code defect in the current audit or hook.
Single-line hotfix: No — there's no code to fix; find and delete the stray logs/ directory under .claude/skills-global/ or .claude/skills/ on the affected machine (or run the audit's own --fix husk-cleanup sweep), then let the streak clear on the next run.
Reduce-complexity option: N/A — constraining scan roots further wouldn't help; they're already correctly scoped. This isn't an audit design flaw, it's leftover state from a bug fixed in #1901/#1907.
The
skills-auditreflection observed a FAIL finding on 2 consecutive runs for skilllogs(rule 19) invalor.Message: Husk directory: no SKILL.md (contains: validate_design_system_sync.jsonl) — delete or restore
This issue was auto-filed by the
skills-auditreflection. It will NOT be re-filed for 30 days even if the finding persists. Close this issue to silence; the streak counter will reset naturally when the underlying rule passes.Triage 2026-09-15
Status: false positive (stale/leftover, not a scan-root bug) — I checked the audit script directly (
.claude/skills-global/audit-skills/scripts/audit_skills.py): rule 19'srootslist (line ~1206-1210) is already correctly scoped to.claude/skills-globaland.claude/skillsonly, via_resolve_repo_root()(lines 47-55) — it never touches repo-rootlogs/. Nologsdirectory exists under either skills root in this checkout today. The root cause was a different, already-fixed bug:.claude/hooks/validators/validate_design_system_sync.pyused to write its JSONL log cwd-relative, so running it with cwd inside a skills root created a stray<skills-root>/logs/husk — fixed in commitc1e7de937("anchor design-system-sync hook log path to repo root (#1901)"), which predates this issue by ~7 weeks and is explicitly documented in a comment atvalidate_design_system_sync.py:53-58. The 2-consecutive-run FAIL this issue was filed from is most likely a leftover directory on whichever machine/checkout the reflection scanned, or a stale cached result — not a live code defect in the current audit or hook.Single-line hotfix: No — there's no code to fix; find and delete the stray
logs/directory under.claude/skills-global/or.claude/skills/on the affected machine (or run the audit's own--fixhusk-cleanup sweep), then let the streak clear on the next run.Reduce-complexity option: N/A — constraining scan roots further wouldn't help; they're already correctly scoped. This isn't an audit design flaw, it's leftover state from a bug fixed in
#1901/#1907.