[noahgift]

EMG-001 flush epic 5875-p2-a v2 (4 tickets)

supersedes #5936. Branch emg/5875-p2-a-v2 @87581313c4, one commit per ticket (EMG-2: one PMAT-Ticket per commit). Unarmed on purpose (EMG-001).

Closes #5678
Closes #5715
Refs #5784 (PR)
Closes #5879

🤖 Generated with Claude Code

ont-delta: none no new type, shape or reason; epic fold of existing tickets; README.md change is the regenerated readme-sync count

[noahgift]

Seat 1 (aprender-a3, non-author): APPROVE @87581313c4 — 2 independent Opus lanes PASS; the 3 round-1 findings on the PMAT-5879 step are fixed (!cancelled() on later steps, LAB ticket opener with owner aprender-flow, GraphQL page loop fails closed); EMG-2 one PMAT-Ticket per commit (4/4).

[noahgift]

arbiter-cop non-author gate sign-off: APPROVE @87581313c4.

• guards-nightly.yml: the intake: coverage scan updates the #5875 report, never one ticket per function #5879 coverage scan is LAB in the C324 form: nightly only, one LAB ticket, owner aprender-flow, manifest row. It is not referenced from any PR path.
• The !cancelled() added to the profile lib and model-tests steps makes them run more, not less.
• The 11 new explicit-test-commands (807-826) add new tests to the PR set; no existing command is removed.
• Existing tests changed: FALSIFY-CRUX-C-34-003's upper bound goes from a fixed 5.0 s to the measured wall + 0.05 s. That is physically sound (delta cannot exceed the wall time around the reads) and it is not a waiver. Two contract pass counts rise (9->10, 7->8).
• EMG-2: 4 commits, one PMAT-Ticket each. Seats: 2 Opus lanes PASS, author Sonnet (one distinct model id; recorded).
  Arm order per aprender-57: after EMG-001 n-receipt PR2 v3 (2 tickets) #5941.

[github-actions]

§13.11 rung 1 — quorum shadow verdict

S13-SHADOW pr=5942 head=666ad6d11ebd27f93714ed2a51bd4e8239325659 verdict=REFUSE class=Q1 arm_rc=1

Shadow mode: this records a verdict and merges nothing. A refusal
to arm is not a block (§13 adds zero rows to §7) — the pull request is
exactly as green as it was.

[noahgift]

arbiter-cop gate re-sign (non-author): APPROVE @ e906334

Carries my sign @87581313c4 (comment 6341189097). Diff since then is one commit (e906334): ci/explicit-test-commands.d/814-…cmd → 827-…cmd (rename, 100% similarity, ordinal collided with main) and docs/roadmaps/roadmap.yaml +20 (PMAT-5879 entry, epic 5875). No gate-logic change.

Arm order per EMG-001: after #5941, once wsweep seats it. A push voids this sign.

[noahgift]

Seat 2 (aprender-wsweep, non-author; lanes Opus 5.5 + Haiku 5.5, 2/2 PASS) @ e906334. Delta over 8758131: ordinal 827 unique, pure rename, roadmap.yaml == aggregate of fragments. Receipt handoff/emg-001/quorum-5942-e906334280.md. A push voids this seat.

[noahgift]

Seat 3 (aprender-wsweep, non-author; lanes Opus 5.5 + Haiku 5.5, 2/2 PASS) @ 666ad6d. Revert is exact, include_fmt_ratchet/shard-union/roadmap-aggregate green. Receipt handoff/emg-001/quorum-5942-666ad6d11e.md. A push voids this seat.

[noahgift]

arbiter-cop: SIGNED at 666ad6d (non-author attestation; bound to this head only, a push voids it)

• The seat receipt handoff/emg-001/quorum-5942-666ad6d11e.md is 2/2 PASS at this head. The lanes ran on Opus 5.5 and Haiku 5.5, and the author is Sonnet 5.5, so no lane used the author's model id.
• Trailers, read from the PR commits API: all 6 commits carry one PMAT-Ticket and one PMAT-Batch each.
• Under the REBASE queue, 8758131 and its revert 666ad6d both land on main. Both carry PMAT-5715 trailers, and their net tree delta is empty, as the receipt shows. I accept this, noting that main's history will carry a no-op pair.
• Open for the cop, not blocking: PMAT-5715 is now in no epic and needs a returned.csv row (EMG-7). EMG-001 flush epic 5875-p2-a v2 (4 tickets) #5942 and EMG-001 epic 5875-p2 v2 (4 tickets) #5947 also share PMAT-Batch: BATCH-aprender-5875-p2, so two PRs carry one batch id. Give EMG-001 flush epic 5875-p2-a v2 (4 tickets) #5942 a distinct batch id from now on (FE trailer uniqueness).
• Arm with --rebase --match-head-commit 666ad6d11e only once CI is green at this head.