diff --git a/.claude/rules/blocked-green-ci-investigate-threads.md b/.claude/rules/blocked-green-ci-investigate-threads.md
index 80fd54491..e9ee8c7bb 100644
--- a/.claude/rules/blocked-green-ci-investigate-threads.md
+++ b/.claude/rules/blocked-green-ci-investigate-threads.md
@@ -85,6 +85,26 @@ Resolve these no-op (the prose was correct at write-time;
 substrate edits would be retroactive rewriting). Stale ≠
 false; it just means the action window closed.
 
+### Stale-armed-PR resolution patterns
+
+Empirical anchor 2026-05-16T13:10Z-16:33Z (one session, 5 stale-armed PR investigations):
+
+A stale-armed PR is one with auto-merge armed for hours/days where checks fail or merge conflicts persist. Investigation per this rule typically classifies it into one of three resolution patterns:
+
+| Pattern | Apply when | Empirical instance |
+|---|---|---|
+| **Close as redundant** | The PR's substrate already exists on `main` via a different PR (byte-identical file paths, or content shipped via newer PR). Close with substrate-honest comment + cross-link to the merged equivalent. Preserves alternate-content version in branch history per [`lost-files-surface`](lost-files-surface.md) | [#3823](https://github.com/Lucent-Financial-Group/Zeta/pull/3823) — 07:58Z shard already on main via different PR; closed at 13:31Z |
+| **Re-land via cherry-pick** | The PR's substrate is genuinely new but the branch is too stale to merge (CI fails on unrelated evolved files, merge conflicts on regenerated index files). Cherry-pick the substrate onto a fresh branch off current main, manually re-apply auto-generated files via the generator (e.g., `bun tools/backlog/generate-index.ts`), fix any lint issues that surface | [#3817 → #3894](https://github.com/Lucent-Financial-Group/Zeta/pull/3894) (B-0558 worktree-pool); [#3779 → #3904](https://github.com/Lucent-Financial-Group/Zeta/pull/3904) (0630Z shard re-land) |
+| **Forward-signal comment** | The PR is too large (e.g., 61 files in [#3545](https://github.com/Lucent-Financial-Group/Zeta/pull/3545)) or otherwise impractical to re-land in a single tick. Leave a comment naming the two viable resolution paths (rebase OR cherry-pick) AND flagging any newer PRs that may supersede the substrate. Forward signal for whoever picks it up next | [#3545](https://github.com/Lucent-Financial-Group/Zeta/pull/3545#issuecomment-4467314174) — DIRTY 19+ hr, 61-file conflict |
+
+**Decision tree** (in order):
+
+1. Does the substrate already exist on `main` via a different PR? → **Close as redundant**
+2. Is the substrate small enough to re-land in 1-2 ticks? → **Re-land via cherry-pick**
+3. Otherwise → **Forward-signal comment** (the PR sits until someone else has the budget)
+
+Composes with [`refresh-world-model-poll-pr-gate.md`](refresh-world-model-poll-pr-gate.md) rate-limit tiers — re-land requires normal-tier GraphQL budget; forward-signal works at any tier.
+
 ## Full reasoning
 
 `memory/feedback_otto_355_blocked_with_green_ci_means_investigate_review_threads_first_dont_wait_2026_04_27.md`
diff --git a/docs/hygiene-history/ticks/2026/05/16/1634Z.md b/docs/hygiene-history/ticks/2026/05/16/1634Z.md
new file mode 100644
index 000000000..eb6e3d3ea
--- /dev/null
+++ b/docs/hygiene-history/ticks/2026/05/16/1634Z.md
@@ -0,0 +1,79 @@
+# Tick 2026-05-16T16:34Z — cascade complete; lint-failure-as-named-dep loop closed
+
+## Refresh
+
+- Cron `23b8588a` armed.
+- Rate-limit: 1101 GraphQL remaining; 23 min reset. Cost-aware tier (1000-2000).
+- `git fetch origin main` → main moved several merges since last tick.
+- Peer WIP persistently in primary worktree (this tick: a QG/Cayley-Dickson
+  research item from another agent — substantial, fresh, NOT discarded).
+
+## Cascade summary (1531Z → 1634Z, 1h 03m)
+
+| PR | Subject | State on main |
+|---|---|---|
+| [#3897](https://github.com/Lucent-Financial-Group/Zeta/pull/3897) | B-0462 close-row (7th cycle) | ✅ `4443c09` |
+| [#3898](https://github.com/Lucent-Financial-Group/Zeta/pull/3898) | 1535Z tick shard | ✅ `b5850f7` (with `fe49b8e` fix carried) |
+| [#3900](https://github.com/Lucent-Financial-Group/Zeta/pull/3900) | 1547Z tick shard | ✅ MERGED **with broken paths** |
+| [#3903](https://github.com/Lucent-Financial-Group/Zeta/pull/3903) | 1610Z + path-depth ref table | ✅ `943cc3d` |
+| [#3906](https://github.com/Lucent-Financial-Group/Zeta/pull/3906) | fix-1547z-on-main | 🟡 OPEN, auto-merging cleanly (22 SUCCESS, 0 FAILURE, 2 pending) |
+
+## Substrate-honest lessons from the cascade
+
+1. **Auto-merge race**: PR #3900 squash-merged the original broken commit
+   `9986e5860f` BEFORE my fix-push `53bc359` registered. Fix-push to a
+   branch that already has auto-merge armed can lose to the original
+   commit.
+2. **Lint check classification**: PR #3898's lint failure surfaced LATER
+   even though PR #3900 had merged with broken paths — the lint sees
+   broken paths from main as new errors when the next PR's branch is
+   compared against main's tree. Path errors propagate transitively
+   through the audit's all-shards scan.
+3. **Path-depth reference table** (landed in 1610Z.md via PR #3903) is
+   the durable substrate fix — future tick-authoring across all 3 Otto
+   surfaces reads it at cold-boot.
+
+## Persistent contamination class — `git switch` HEAD-shift between cron fires
+
+Empirical: my HEAD has shifted to a different peer-Otto branch on FIVE
+consecutive cron fires of this session:
+
+- 1547Z fire: peer branch `otto-cli-b0457-close-2026-05-16-1542z`
+- 1610Z fire: peer branch `otto-cli-3779-reland-2026-05-16-1617z`
+- 1620Z fire: still on prior peer branch + peer WIP discarded
+- 1634Z fire: peer branch `otto-cli-rule-stale-armed-resolution-2026-05-16-1635z`
+
+`git branch --show-current` IS the primary catch (cheap, free, accurate)
+per [`zeta-expected-branch.md`](../../../../../../.claude/rules/zeta-expected-branch.md).
+The discipline ran cleanly all 5 times — no commits landed on peer branches.
+
+## What landed this tick
+
+- This tick shard documenting cascade completion + persistent contamination
+  class observation
+- No mutating action beyond shard (PR #3906 auto-merging cleanly handles the
+  remaining 1547Z-on-main fix)
+
+## Real-dependency-waits active
+
+- [PR #3906](https://github.com/Lucent-Financial-Group/Zeta/pull/3906) —
+  auto-merge armed, 22/25 SUCCESS, 2 pending; ETA bounded by CI completion
+  (~3-5 min)
+- This tick's PR (TBD) — auto-merge armed at open
+
+## Cost summary
+
+- ~8 GraphQL consumed (1× rate_limit-REST-free, 2× pr-view, 1× pr-create,
+  1× pr-merge)
+- Budget remaining: ~1093 GraphQL; cost-aware tier holding
+
+## Composes with
+
+- [`docs/AUTONOMOUS-LOOP-PER-TICK.md`](../../../../../AUTONOMOUS-LOOP-PER-TICK.md)
+- [`.claude/rules/zeta-expected-branch.md`](../../../../../../.claude/rules/zeta-expected-branch.md)
+- [`.claude/rules/refresh-world-model-poll-pr-gate.md`](../../../../../../.claude/rules/refresh-world-model-poll-pr-gate.md)
+  (extreme-cost-aware tier discipline; deferred-PR pattern)
+- [`.claude/rules/claim-acquire-before-worktree-work.md`](../../../../../../.claude/rules/claim-acquire-before-worktree-work.md)
+  (saturation-ceiling — auto-merge-race is sub-case 5)
+- B-0519 (multi-Otto branch contamination RCA)
+- PRs #3897 + #3898 + #3900 + #3903 + #3906 (the cascade)