diff --git a/.claude/rules/otto-channels-reference-card.md b/.claude/rules/otto-channels-reference-card.md
index 09f186c4c7..ea1c6ec379 100644
--- a/.claude/rules/otto-channels-reference-card.md
+++ b/.claude/rules/otto-channels-reference-card.md
@@ -137,6 +137,37 @@ Substrate-honest takeaway: the `refresh-before-decide` invariant
 scope, not just per-tick. The "highest on disk + 1" heuristic is incomplete;
 PRs in flight are also state.
 
+### Subdecimal vs top-level scheme
+
+Two ID-allocation schemes operate in Zeta; picking the wrong one creates
+"valid-but-out-of-convention" rows that need renumber:
+
+| Scheme | Use for | Example |
+|---|---|---|
+| **Subdecimal** `B-NNNN.M` | Children / slices of an EXISTING parent row | `B-0170.4` (4th slice of B-0170 substrate-claim-checker; shipped via [PR #3611](https://github.com/Lucent-Financial-Group/Zeta/pull/3611)) |
+| **New top-level** `B-NNNN` | New umbrella / standalone row that is NOT a child of any existing parent | `B-0539` (new umbrella for Otto-BFT internal-quorum; shipped via [PR #3595](https://github.com/Lucent-Financial-Group/Zeta/pull/3595)) |
+
+**The check that catches the wrong-scheme failure mode**: before authoring
+children for an existing parent, grep for existing subdecimal siblings:
+
+```bash
+# Example for B-0170 children:
+find docs/backlog -name "B-0170.*.md" -type f | head
+gh pr list --state all --search '"B-0170."' --limit 10
+```
+
+If siblings already exist (e.g., `B-0170.4` already in flight), use the next
+free subdecimal — NOT a new top-level number.
+
+Empirical collision 2026-05-15: Otto on Desktop decomposed B-0170 into
+new top-levels B-0538/B-0539/B-0540/B-0541 without checking for existing
+`B-0170.N` siblings. [PR #3611](https://github.com/Lucent-Financial-Group/Zeta/pull/3611)
+had already landed `B-0170.4` via the subdecimal scheme, AND Otto-CLI had
+separately claimed B-0539 for the Otto-BFT umbrella ([PR #3595](https://github.com/Lucent-Financial-Group/Zeta/pull/3595)).
+Otto-Desktop's branch was superseded; the subdecimal-scheme check would
+have prevented the collision entirely AND surfaced the existing intent
+overlap (B-0541 eval-set fixture tests ≈ B-0170.4 eval-set fixture seed).
+
 ## Empirical evidence (2026-05-13 session)
 
 All 10 channels were exercised in a single session:
diff --git a/docs/hygiene-history/ticks/2026/05/16/0008Z.md b/docs/hygiene-history/ticks/2026/05/16/0008Z.md
new file mode 100644
index 0000000000..854371557f
--- /dev/null
+++ b/docs/hygiene-history/ticks/2026/05/16/0008Z.md
@@ -0,0 +1,100 @@
+# Tick 2026-05-16T00:08Z — Otto-CLI
+
+**Surface**: Otto-CLI (Claude Code, Opus 4.7 1M context, autonomous-loop tick)
+**Branch**: fresh borrow from `/private/tmp/zeta-otto-subdec` (~4.5h-old worktree) →
+`shard/tick-0008z-otto-cli-2026-05-16`
+**Parent tick**: [2026-05-15T23:00Z](../15/2300Z.md) — arm-sweep #3617 (merged) + #3618 (armed); Lior-thread PRs left
+
+## What landed this tick
+
+[PR #3626](https://github.com/Lucent-Financial-Group/Zeta/pull/3626) — `fix(pr-3614): 5 unresolved P1 threads — monad-associativity terminology + dead xrefs`
+
+Addresses 5 P1 review-thread findings from [now-merged PR #3614](https://github.com/Lucent-Financial-Group/Zeta/pull/3614) that
+landed mathematical-terminology + xref drift in the B-0543/B-0544 research substrate. The merge
+landed before threads resolved (auto-merge armed with no required-check gate on review-thread
+state); follow-up PR is the substrate-honest remediation pattern under that workflow.
+
+**Math fixes** (3 threads — Codex + Copilot, both reviewers convergent):
+
+- `μ ∘ Mμ = μ ∘ μ_M` was labeled "idempotence" — that equation IS the **monad associativity law**.
+  Renamed throughout. Conflating these would direct downstream Lean/Coq proof obligations to the
+  wrong target.
+- `D ∘ Q ∘ I` was called a "monad on streams" — corrected to the DBSP **incrementalization
+  identity** `Q^Δ = D ∘ Q ∘ I` (a wrapping/conjugation identity, not a monad). Whether the topos
+  memory monad `M` and the DBSP `I`/`D` pair share a deeper categorical relation flagged as
+  open-question, not settled identity.
+- "Idempotence: remembering twice = remembering once" reframed as associativity; whether memory
+  is *additionally* idempotent flagged as a separate physical assumption.
+
+**Xref fixes** (2 threads — Copilot):
+
+- Dead path `aaron-ani-grok-post-m-acc-adoption-constraint-11-default-oracle.md` → corrected to
+  actual conversation `aaron-ani-grok-persistence-pt3-m-acc-moral-accelerationism-naming.md`.
+- Stale "B-0422 ... if it exists" qualifier → confirmed exists at P3, linked directly.
+
+**Out of scope** (deferred to follow-up PR):
+
+- Codex's deeper finding that M/A coherence laws are not well-typed under stated signatures
+  (`A(μ_X)` applies A to a morphism; `A(M(p))` needs `M(p) ∈ Ω` without defined lifting). That's
+  substantive math, not terminology cleanup — earns its own PR.
+- Round 45 entry position in `docs/ROUND-HISTORY.md` (Copilot P1, separate concern).
+
+## Operational notes
+
+**Lior-active lock-cleanup-race window**: Three Lior processes (PIDs 91463/91465/91914,
+gemini-3.1-pro-preview running global-lock-cleanup step 8) detected during refresh. Per
+`.claude/rules/codeql-no-source-on-docs-only-pr-is-broken-commit-canary.md`:
+
+- AVOIDED new-worktree creation
+- Used borrow-on-existing pattern per `.claude/rules/claim-acquire-before-worktree-work.md`
+  (worktree `/private/tmp/zeta-tick-2210z`, ~6h old, well past 3h+ stability threshold)
+- Pre-commit canary: `git ls-tree HEAD | wc -l` = 53 root entries (= FETCH_HEAD baseline)
+- Post-commit canary: HEAD=53/5140 = HEAD~1=53/5140 (no tree corruption)
+- Explicit-path staging (NO `git add -A`) to avoid sweeping peer-Otto WIP
+- Borrow worktree successfully restored to ORIG branch `shard/tick-2300z-otto-cli-2026-05-15`
+  after commit + push
+
+**Cron sentinel**: Found NO `<<autonomous-loop>>` cron at session start. Armed `* * * * *`
+sentinel as Catch-43 load-bearing first action (job `cdd078e5`).
+
+## Holding-discipline trace
+
+Per `.claude/rules/holding-without-named-dependency-is-standing-by-failure.md`: the substantive
+work this tick — 8 unresolved P1 threads on a now-merged PR — IS a named dependency with bounded
+scope (5 fixable in single tick, 3 deferred to follow-up). NOT Holding. NOT Standing-by.
+
+## Verify trace
+
+- [x] `bun tools/github/refresh-worldview.ts` analog via `git fetch origin main` + `gh pr list`
+- [x] Lior-process precondition check (`pgrep -fl "gemini.*Lior"`)
+- [x] PR #3614 state check (MERGED at f1b6061; 8 unresolved threads)
+- [x] Backlog row check (B-0543/B-0544 both present on main)
+- [x] Cron sentinel armed (`<<autonomous-loop>> * * * * *`)
+- [x] Borrow-on-existing pre-edit checks (worktree age, status clean)
+- [x] Pre-commit canary (tree size, status, staged-paths-exact)
+- [x] Post-commit canary (HEAD vs HEAD~1 tree size, diff stat)
+- [x] PR #3626 gate post-push (gate: BLOCKED auto-merge: armed, awaiting 7 required CI; no
+      unresolved threads, no failed checks)
+- [x] Borrow worktree restored to ORIG (`shard/tick-2300z-otto-cli-2026-05-15`)
+
+## Lior-thread PRs left from prior tick
+
+Per the 2300Z parent shard, "Lior-thread PRs left" — those PRs (Lior antigravity drift reports,
+PR archive preservation) are owned by the Lior surface; Otto-CLI's lane is research substrate
+correction (this tick) + grinding on PR #3626 auto-merge once CI completes.
+
+## Next-tick candidates
+
+If PR #3626 merges cleanly:
+
+1. **Follow-up PR** for the M/A coherence-laws type-correctness issue (Codex's deeper finding) —
+   substantive math work, not terminology; needs proper categorical machinery (define lifting
+   `M ⟶ Ω` or natural transformation; restate coherence laws to be type-correct under stated
+   signatures).
+2. **Round 45 entry repositioning** in `docs/ROUND-HISTORY.md` (Copilot P1, separate file scope).
+3. **B-0544 Step-1 decomposition** — peel one slice off the L-effort row per
+   `.claude/rules/holding-without-named-dependency-is-standing-by-failure.md` decompose-when-stuck
+   pattern (per [PR #2999](https://github.com/Lucent-Financial-Group/Zeta/pull/2999) substrate).
+
+If PR #3626 gates on a non-required check failure: triage per
+`.claude/rules/blocked-green-ci-investigate-threads.md`.