shard(0816Z): Lint-frontier closure + auto-merge-race lesson

docs/hygiene-history/ticks/2026/05/19/0816Z.md

@@ -0,0 +1,63 @@
+# Tick 0816Z — 2026-05-19 — Lint-frontier closure + auto-merge-race lesson
+
+## Cold-boot context
+
+Cycle-new after PR #4357 merged at `cfba8a64` (sentinel `02c996c5` alive across the cycle). Brief-ack counter reset by named-dep #1 (own-PR merge).
+
+## Step 1 — Refresh
+
+| Signal | Value |
+|---|---|
+| UTC | 2026-05-19T08:16:58Z |
+| `origin/main` tip | `cfba8a64` (PR #4357 — the 0803Z shard I shipped previous tick) |
+| GraphQL remaining (pre-spend) | 282 → Extreme cost-aware (200-1000) per [`refresh-world-model-poll-pr-gate.md`](../../../../../../.claude/rules/refresh-world-model-poll-pr-gate.md) |
+| Peer activity | 14 procs (similar to 0803Z) |
+| Lint state | 10 NEW findings on `0608Z.md` (PR #4343 inherited) + 3 NEW on `0803Z.md` (my own; auto-merge-raced) |
+
+## Step 2 — Holding discipline
+
+Cycle-new; no brief-ack chain. Substantive work available (lint frontier).
+
+## Step 3 — Pick speculative work
+
+**Substrate-engineering observation**: the `audit-tick-shard-relative-paths --enforce --baseline` lint has 13 NEW findings (10 on `0608Z.md` from PR #4343, 3 on `0803Z.md` from PR #4357). Both PRs merged despite the failures because the check is non-required. Each new tick shard inherits the failing state, which dilutes the signal-value of the lint check.
+
+Closing the frontier costs ~13 character edits across 2 files. Single PR drop.
+
+## Step 4 — Verify + commit
+
+Combined fix in PR #4358 ([`fix/0608z-tick-shard-relative-path-depth-2026-05-19`](https://github.com/Lucent-Financial-Group/Zeta/pull/4358)):
+
+- `docs/hygiene-history/ticks/2026/05/19/0608Z.md` — 10 path corrections (`../../../../../` → `../../../../../../`)
+- `docs/hygiene-history/ticks/2026/05/19/0803Z.md` — 3 path corrections (same pattern)
+- Audit re-run: `exit=0`, `0 NEW findings` (12 grandfathered baseline unchanged)
+- ls-tree HEAD = 53 (matches `origin/main cfba8a64`; no canary corruption)
+
+## Auto-merge race — substrate-honest lesson
+
+PR #4357's first commit `45128146` was armed for auto-merge while CI was still in-progress. My follow-up lint-fix commit `9c9c8e69` pushed to the same branch ~5 min later. The CI then completed, the gate went CLEAN (lint non-required), and auto-merge fired on the FIRST commit's content — my fix never made it onto main.
+
+**The lesson**: auto-merge + non-required-failing-check + follow-up commit is a race window. Either:
+
+1. Don't arm auto-merge until all desired commits are pushed, OR
+2. Treat non-required failures as if they were required when shipping a single-tick fix
+
+Sharpening candidate for [`blocked-green-ci-investigate-threads.md`](../../../../../../.claude/rules/blocked-green-ci-investigate-threads.md): "auto-merge-arms-then-fix" anti-pattern when a non-required lint is failing on the initial commit. Future tick may convert this into a rule-extension PR.
+
+## Step 5 — Shard
+
+This file IS Step 5.
+
+## Step 6 — CronList
+
+Sentinel `02c996c5` alive (`* * * * *` recurring `<<autonomous-loop>>`).
+
+## Step 7 — Visibility-stop
+
+Concrete artifact: PR #4358 (combined 0608Z+0803Z lint fix; auto-merge armed) + this tick shard.
+
+Counter cycle: this tick is concrete-artifact (condition #3); cycle-new closes cleanly.
+
+GraphQL post-spend: ~51 → **Pure-git tier**. This shard committed + pushed via pure-git per the [`refresh-world-model-poll-pr-gate.md`](../../../../../../.claude/rules/refresh-world-model-poll-pr-gate.md) tier table; PR creation for the shard branch deferred to post-reset tick.
+
+End of tick.