shard(tick): 2026-05-16T02:58Z — GraphQL reset → arm wave + 8-thread cleanup

docs/hygiene-history/ticks/2026/05/16/0258Z.md

@@ -0,0 +1,85 @@
+# Tick 2026-05-16T02:58Z — Otto-CLI
+
+**Surface**: Otto-CLI (Claude Code, Opus 4.7 1M context, autonomous-loop tick)
+**Parent tick**: [2026-05-16T02:52Z](0252Z.md) — audit filter triage (17→10)
+
+## What landed this tick
+
+GraphQL reset (02:55:28Z) → arm + thread-cleanup wave. Three substantive fixups + one peer-merge observation.
+
+| PR | Action | Result |
+|----|--------|--------|
+| [#3690](https://github.com/Lucent-Financial-Group/Zeta/pull/3690) | Auto-merge armed; 1 Copilot thread (table-pipes) verified **false positive** (single-`\|` tables, no double-pipe) → resolved no-op | OPEN, armed, 0 unresolved |
+| [#3692](https://github.com/Lucent-Financial-Group/Zeta/pull/3692) | Auto-merge armed; 4 Copilot threads (sonarjs disable, `main` export + `import.meta.main` guard, generic URI scheme regex, `--files` validation) — **all real**; fixup commit `d299c7d` pushed; threads resolved | OPEN, armed, 0 unresolved |
+| ~~[#3693](https://github.com/Lucent-Financial-Group/Zeta/pull/3693)~~ | Found MERGED at 02:55:44Z by AceHack 16s after GraphQL reset; clean | MERGED |
+| [#3695](https://github.com/Lucent-Financial-Group/Zeta/pull/3695) | Auto-merge armed; 3 Copilot threads — 2 real prose errors (wrong rule citation; wrong TTL math) fixed via commit `e0828b5`; 1 stale (parent-tick link to 0249Z.md, now on main via merged #3693) resolved no-op | OPEN, armed, 0 unresolved |
+
+## What the 8 threads revealed
+
+### Audit-script PR #3692: all 4 findings were real and high-quality
+
+Copilot caught the standard repo conventions I'd missed:
+
+1. **`spawnSync("git", ...)` needs `// eslint-disable-next-line sonarjs/no-os-command-from-path`** — every sibling tool that shells out to git uses this exact comment; the lint gate would fail without it.
+2. **Top-level `process.exit(main(...))` blocks module-import** — sibling `audit-section-33-migration-xrefs.ts:291` uses `if (import.meta.main) { process.exit(main(...)); }` as the test-safe pattern.
+3. **`isRelativeTarget` was too narrow** — only excluded `http(s)` and `mailto`; missed `ftp:`, `file:`, `tel:`, `data:`. Replaced with generic `/^[A-Za-z][A-Za-z0-9+.-]*:/` scheme regex.
+4. **`--files` inputs not validated** — `readFileSync` throws on missing path, bypassing the documented exit-64 contract. Added existence check at args boundary.
+
+All 4 caught by Copilot before merge — net positive for the audit's robustness.
+
+### Shard PR #3695: 2 real prose errors in my tick-8 shard
+
+- I cited "`claim-acquire-before-worktree-work.md` rule's 'ID allocation discipline' section" — that section is actually in `otto-channels-reference-card.md`. Replaced with the canonical immutability statement at `docs/hygiene-history/ticks/README.md` ("Each shard is an immutable per-tick event" per Git-native CQRS / Event Sourcing framing).
+- I wrote "check at 25:44 from then = 02:44Z+" for the bus-claim TTL — that's wrong math. 24h from 01:44Z is 2026-05-17T01:44Z, not same-day 02:44Z. Fixed.
+
+### Shard PR #3690: table-pipe complaint was Copilot's repeated false-positive
+
+`awk` line-by-line check confirmed lines 18, 37, 65 all use single `|` (clean 2-col and 4-col tables). Same hallucination pattern Copilot exhibited on PR #3685 in tick 6. Resolved no-op.
+
+## State at tick end
+
+3 PRs in queue, all armed + thread-clean, awaiting CI:
+
+- **PR #3690** (0240Z shard from tick 6) — pre-existing un-armed, now armed
+- **PR #3692** (audit script + filter + 4 quality fixups) — armed
+- **PR #3695** (0252Z shard from tick 8 + prose fixups) — armed
+
+GraphQL: 4564 remaining after the burst (started tick 9 with 0 → 5000 at 02:55:28Z reset → 4564 after arming + thread queries + mutations). Well within budget for future ticks.
+
+## Discipline reinforced
+
+### "Verify before fixing" — Copilot reviewer-feedback edition
+
+Two distinct false-positive patterns this session:
+
+1. Table-pipe `\|\|` hallucination — appeared on PR #3685 (tick 6) and PR #3690 (tick 9); verified false on both by direct `awk` line inspection
+2. `0249Z.md not present on branch` — was correct AT REVIEW TIME but resolved itself when sibling-PR-merged-to-main (PR #3693)
+
+Direct inspection beats taking reviewer output at face value. Same shape as `verify-before-deferring.md` extended to reviewer feedback. Applied this tick on both #3690 (no edit needed) and the audit-script quality findings (all 4 verified real before fixing).
+
+### "Audit ships in iterations" — the 4-step pattern at compressed cadence
+
+PR #3692's lifecycle this session:
+
+| Step | Tick | Commit |
+|------|------|--------|
+| 1. Discovery (the bug class shipped in #3676 + #3679) | 3 | n/a |
+| 2. Narrow fix (per-shard link fix) | 3 | #3676 fixup + #3680 |
+| 3. Scanner authored | 7 | initial PR #3692 |
+| 4. Filter improvement | 8 | fixup `e945016` |
+| 5. Quality findings cleared | 9 | fixup `d299c7d` |
+| 6. Baseline cleanup (TBD) | next | open question |
+| 7. CI enforce (TBD) | post-baseline | open |
+
+Same shape as §33 audit's lifecycle (PR #3513 → #3529 → #3548 → #3552 → enforce) but compressed across ticks instead of across weeks.
+
+## Holding-discipline state
+
+Concrete code substrate (audit-quality fixes) + prose substrate (shard cross-reference corrections) landed this tick. Counter resets per "actually picking real decomposition work" clause.
+
+## Next-tick candidates
+
+1. **Monitor the 3 armed PRs to merge** + investigate any new threads
+2. **Baseline-cleanup decision** for the 10 remaining audit findings (strict-vs-pragmatic immutability reading; deferred since tick 8)
+3. **B-0545 renumber-sweep** (claim active until 2026-05-17T01:44Z)
+4. **CI gate** for the audit, contingent on baseline-cleanup decision