G87-D5 Band-C and gate-84 handoff classification

[briansrls]

Auto-opened by session-dashboard for session proud-koi-122.
Pushing to session/proud-koi-122 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

• Title describes the change (not the session id or branch).
• PR body summarises what and why (replace the TODO below).
• Tests run: name the command (e.g. npm test, cargo test) and the result.
• If this closes a work item, the body contains a Closes #N directive.
• No commits on this branch are surprises (no fork/cherry-pick I did not make).
• No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

• TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

[briansrls]

Re-verified the codex review against HEAD (9cd3c5f): the four src/v3/compiler/tests/integration/cementing/*.rs entries in EXPECTED_HAND_AUTHORED_TEST match the §3 table, and the G87-D5 four-way taxonomy matches r3-gate-87-lens-cementing-test-discipline-dispatch-2026-05-13.md §G87-D5. No further doc or code edits indicated.

dashboard-ops reviews gunb-ai/gunbc#2952 still shows 1/2 distinct dashboard approvals (codex only), so squash-merge stays deferred until the second provider lands; mergeable=CLEAN and checks passing otherwise.

— sent from proud-koi-122

[briansrls]

Review metadata

• Provider / model: openai-pro / gpt-5-5-pro
• Commit: 9cd3c5f9 · Trigger: manual
• Comparison: main @ c961e667 ... session/proud-koi-122 @ 9cd3c5f9
• Conversation: View conversation

---

1. Story of the diff

This PR is a docs-only handoff clarification for Gate #87 / #84 cementing work. The previous brief table treated the four tests/integration/cementing/*.rs residuals mainly as predicate/blocker rows; the new text first defines an explicit G87-D5 dispatch taxonomy at docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:72-77, then reclassifies each cementing residual into that taxonomy at docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:85-88. The load-bearing effect is routing discipline: the brief now says none of the four cementing integration modules are Band-C bulk-port candidates at HEAD, so #84 workers should not pick from this subdirectory for bulk-port pilots and should instead use the coordinator live inventory for other SG-0 hand-authored-test rows (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:81). The row-level blockers and SG-0 deltas remain explicit, so the brief is narrowing worker selection without pretending any Rust residual has already been dissolved.

2. Invariant categories

1. LAYER MODEL (substrate vs implementation).

N/A — the diff is a planning-brief/documentation change only; it does not touch Dag, substrate declarations, compiler types, lowering/inference/emission code, or runtime mutation. The new material is a dispatch taxonomy and table for existing hand-authored integration-test residuals (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:72-88).

1. INVARIANTS.md + modeling-discipline.md.

Compliant — P5 / Progress Is Dissolution is handled by keeping each residual tied to a blocker, owning lane, and dissolution/census outcome instead of turning the four Rust tests into permanent scaffolding: provenance has a named expected-carrier blocker and -1 trigger (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:85), complexity has Gate-73 carrier unblock and -1 trigger (...:86), cost is explicitly removed from Band-C cementing and routed to gate #78 host-wrapper retirement (...:87), and memory-peak is routed to T-LAS gates #91/#94 with a -1 trigger (...:88). That matches the invariant that scaffolds need explicit dissolution triggers, not steady-state bridge status. chatgpt-review-73df3d6d-1c64-49…

1. CODING.md.

N/A — no Rust implementation code, helpers, methods, error/result shapes, or APIs are added. The CODING.md data/free-function discipline is therefore not exercised by this diff. chatgpt-review-bc61f5ee-3367-4f…

1. TESTING.md.

Compliant — no new test is required for this docs-only classification change because it does not promote a lens row to COMPLETE, add a new Rust receipt, or claim a port occurred. It explicitly says the brief “makes no census-delta claim” (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:79) and that no cementing-row bulk-port candidate exists at HEAD (...:81). That is consistent with the cementing-test rule that temporary Rust receipts are allowed only with named blockers and remain in EXPECTED_HAND_AUTHORED_TEST until a .dag harness replaces them. chatgpt-review-4a52fe3b-f347-4a…

1. LOCKED DESIGN DECISIONS.

N/A — the diff does not alter Pure Bootstrap, substrate shape, test-runner design, or any locked design authority. It routes work under existing zero-floor / cementing discipline rather than changing those decisions.

1. TRACKED vs UNTRACKED DEBT.

Compliant — the PR is specifically making the debt more tracked: it introduces a mutually exclusive category set for remaining cementing/*.rs rows (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:72-77), records that the four-row census was rechecked (...:79), and says there are no Band-C bulk-port candidates in this subdirectory (...:81). Each table row preserves documentation, bounds, and a dissolution trigger or non-delta reason (...:85-88).

2.5. Top-down PM intent review

Compliant — the highest-level intent is still “Rust residuals shrink toward zero; hand-authored tests are temporary carriers for missing .dag predicate/test surfaces, not permanent implementation.” Pure Bootstrap to Zero sets the goal as zero hand-authored v3 source files and commits to shape γ with floor 0, while the acceptance gates require EXPECTED_HAND_AUTHORED_TEST = 0; the thesis likewise says hand-maintained test surfaces shrink toward zero and tests become .dag TestClaim data. chatgpt-review-868e9812-36fc-43…

chatgpt-review-868e9812-36fc-43…

chatgpt-review-8a8f2901-7c72-43…

The diff does not dilute that into “these four Rust files stay forever.” It says no port happened in this PR (docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md:79), prevents #84 workers from misclassifying the cementing subdirectory as bulk-port inventory (...:81), and keeps each residual attached to the blocker that must clear before the Rust row dissolves (...:85-88). That is faithful to the PM intent: it reduces dispatch ambiguity without relaxing the zero-residual target.

3. Verdict

APPROVE. No findings: this is a narrow docs-only routing correction that strengthens, rather than weakens, the handoff classification for Gate #87 / #84 work. The important safeguards are present: no false census-delta claim, no new permanent Rust residual, and every remaining cementing row keeps an owner/blocker plus a dissolution path.

[briansrls]

Verified the openai-pro narrative against current docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md on 9cd3c5f: the G87-D5 four-way taxonomy is at L72–L77, the “no census-delta” / re-check sentence at L79, the bulk-port routing note at L81, and the four-row disposition table at L83–L88 (the review’s “85–88” is off by two because of the new header row and column). Substance matches: taxonomy → census note → bulk-port guardrail → per-row blockers/deltas unchanged in meaning from the pre-G87-D5 table.

No code or doc amendment is required from this review: it is APPROVE with zero requested edits.

Merge gate: dashboard-ops reviews gunb-ai/gunbc#2952 still reports 1/2 parsed dashboard approvals because the openai-pro run is ingested as verdict: "unknown" from the stdout artifact even though the posted GitHub api-review comment ends in APPROVE. Until the dashboard ingest recognizes that verdict (or an operator merges on the human-readable approvals), automated “ready” stays false despite mergeable=MERGEABLE and checks passing.

— sent from proud-koi-122

[briansrls]

Closing — cycle 26 respawn-loop on closed gate-87 scope. — sent from swift-deer-459

[briansrls]

Review metadata

• Provider / model: codex / unknown
• Commit: 9cd3c5f9 · Trigger: schedule
• Thinking: 210s wall

---

Non-blocking — Strengths

• docs/briefs/r3-cementing-discipline-pattern-2026-05-12.md The G87-D5 classifications preserve the gate #87 registry boundary, name the owning lanes for remaining Rust receipts, and keep #84 bulk-port selection tied to the live SG-0 census.

ROADMAP — Verified

• G87-D5: The four cementing/*.rs rows in the table match EXPECTED_HAND_AUTHORED_TEST, on-disk modules, and tests/integration.rs wiring.

✅ No blocking concerns.