shard(tick): 2108Z — review-thread triage round 2 (B-0329 renumber + gh api graphql fix)

docs/hygiene-history/ticks/2026/05/14/2108Z.md

@@ -0,0 +1,103 @@
+# Tick 2026-05-14T21:08Z — review-thread triage round 2 (B-0329 ID-collision renumber + gh api graphql usage fix)
+
+## Refresh (step 1)
+
+- Cron sentinel `12fb713e` live.
+- #3248 (shard 2103Z) merged as `9de7c6e` during this tick.
+- 4 in-flight PRs polled (#3245 / #3246 / #3247 / #3248):
+  - **#3246** (shard 2059Z) — 1 unresolved Copilot thread
+  - **#3247** (BACKLOG regen) — 1 unresolved Copilot thread (substantive)
+  - **#3245 / #3248** — wait-ci, autoMerge armed
+
+## Holding discipline (step 2)
+
+Per `blocked-green-ci-investigate-threads.md`: investigate threads first
+when BLOCKED + autoMerge armed.
+
+## Work (step 3) — two thread fixes, one substantive
+
+### Thread on #3247 — substantive: duplicate B-0329 ID
+
+Copilot catch: "BACKLOG.md adds a second visible B-0329 entry even
+though backlog IDs are required to be factory-wide unique
+(tools/backlog/README.md:61). Renumber or otherwise fix the per-row
+file before regenerating the index."
+
+Real data-integrity bug. Two on-disk files both claimed `id: B-0329`:
+
+- `B-0329-claude-md-as-process-not-doctrine.md` (existing — keep)
+- `B-0329-new-surface-audit-alignment-check.md` (NEW — collision)
+
+Fix:
+
+- Renamed file: `B-0329-new-surface-audit-alignment-check.md` →
+  `B-0520-new-surface-audit-alignment-check.md`
+- Frontmatter `id: B-0329` → `id: B-0520`
+- Body heading `# B-0329 — ...` → `# B-0520 — ...`
+- Added `renumbered_from` breadcrumb citing the Copilot catch + #3247
+- Regenerated BACKLOG.md — duplicate B-0329 row removed, B-0520 picked up
+
+Verifies clean: `grep -c "B-0329" docs/BACKLOG.md` returns 1. No other
+refs to the renamed file (grep confirmed).
+
+Pushed commit `4407e4a` to #3247's branch; thread resolved with reply.
+
+### Thread on #3246 — accuracy nit: gh api graphql usage
+
+Copilot catch: "`gh api graphql resolveReviewThread` isn't a valid
+`gh api graphql` invocation; `resolveReviewThread` is a GraphQL
+mutation field and needs to be passed via `-f query=...`."
+
+Corrected the 2059Z shard's narrative text to show the canonical
+runnable pattern:
+
+```text
+gh api graphql -f query='mutation { resolveReviewThread(input:{threadId:"<id>"}) { thread { isResolved } } }'
+```
+
+Pushed commit `a5b812b` to #3246's branch; thread resolved.
+
+## Verify (step 4)
+
+- B-0329 collision: `grep -c "B-0329" docs/BACKLOG.md` returns 1 (only
+  the original)
+- B-0520 appears at end of P1 section in regenerated index
+- File rename detected by git (`R` status; 85% similarity)
+- Both threads resolved (verified `isResolved=true` via GraphQL)
+- Composite branch-guard used at all commits
+
+## Shard (step 5)
+
+This file.
+
+## CronList (step 6)
+
+Sentinel `12fb713e` armed; one entry, recurring.
+
+## Visibility (step 7)
+
+- **PR #3247** (BACKLOG regen): commit `4407e4a` pushed (renumber +
+  regen); thread resolved; CI re-running
+- **PR #3246** (shard 2059Z): commit `a5b812b` pushed (gh api usage
+  fix); thread resolved; CI re-running
+- **#3245 / #3248** still wait-ci, autoMerge armed
+- **#3248** (shard 2103Z) merged as `9de7c6e` during the tick
+
+## Notes for future-Otto
+
+ID-collision class of bugs (multiple on-disk files claiming the same
+`id: B-NNNN`) is a real data-integrity failure mode. The
+`tools/backlog/generate-index.ts` output makes it visible (duplicate
+checkbox lines) but doesn't refuse to generate; Copilot's catch was the
+review-time signal.
+
+The `renumbered_from` frontmatter breadcrumb pattern preserves the
+archaeology without needing a separate ADR — anyone grep-ing for the
+old ID in the renamed file finds the breadcrumb explaining why.
+
+Audit-discipline candidate (filed mentally for next tick): a small
+hygiene script `tools/hygiene/audit-backlog-id-uniqueness.ts` that
+walks per-row files, parses frontmatter `id` fields, and reports any
+duplicates. Would have caught this at author-time rather than review-
+time. Not authoring this tick (B-0488 row noted similar tools exist
+under `tools/hygiene/`; should check before duplicating).