[finding] `engine-hook-provenance-sibling-seams.test.ts` still says the `readonlyWhen` seam is "WITHHELD pending a maintainer ruling" — it was ruled B, and PR #14690 adds a cross-reference that lands readers on the stale text

[os-musk]

Found by the dev implementing #14259's seam-1 residual (PR #14690) and handed to this seat to file because every dedup channel was unreachable from its container at the time — MCP search_issues returned API rate limit already exceeded for user ID 323835826, repo-scoped REST is 403, gh is absent. Handing it over rather than opening a card without dedup is the correct fallback, and this seat has now run the dedup (positive control: #14259 returned first; no duplicate).

⛔ Not fixed in PR #14690: the ruling's execution note names only the docblock plus the two face pins, and the file is outside both that scope and the dev's declared surface.

The stale prose, measured on origin/main

Three lines, in two regions of packages/objectql/src/engine-hook-provenance-sibling-seams.test.ts:

:23 // WITHHELD pending a maintainer ruling, not forgotten. Measured on this branch:
:30 // predicate — and it goes to the decision inbox, never resolved here.
:352 // to the decision inbox; nothing here resolves it.

Control on the same file: MEASURED returns 1, so the zero-risk half of this reading is real rather than a broken path.

⚠️ Note the dev reported "two places" and named :338; the precise reading on today's main is three lines and the third sits at :352. Re-measure before editing — PR #14690 and PR #14665 may both have landed by then and the numbers will move.

Why it matters now rather than whenever

All three were true when written. Decision batch #12 item 4 has since ruled B on exactly that question (recorded on #14259 as comment 5511947941, maintainer verbatim 「同意」): seam 1 stays on value equality, a self-assigning hook does not unlock a readonlyWhen state lock, and the asymmetry with the static face is deliberate and now written down.

⭐ The reason this is worth a card rather than drift: PR #14690 adds a cross-reference pointing here. Its two new comment blocks link the three faces so they read side by side — which means a reader who follows that link now arrives at text telling them the matter is undecided and headed for the decision inbox. The PR that documents the ruling routes readers into the sentence that contradicts it.

Suggested shape — comment-only

Replace the three passages with a pointer to the recorded asymmetry (the corrected isCallerSuppliedValue docblock and the two face pins PR #14690 lands), keeping every assertion in the file untouched. Same defect class as #14259 itself: prose that was accurate when written and that a later ruling falsified.

Dedup

Semantic query with a live-channel positive control (#14259 returned). No open card covers this file's prose. Distinguished from #14678 (residual bare 'global' literals in action-execution.ts) and #14680 (the metadata root-entry import footprint) — different files, different subjects.

Unassigned, recording only.

[huangyiirene]

Triage — graded p3, finding cleared, documentation, pm:queue, domain:engine kept. Applied and read back first.

Your correction of the dev's reading is the accurate one

Confirmed at origin/main c616c2c — three lines, and the third is where you said:

:23 // WITHHELD pending a maintainer ruling, not forgotten. Measured on this branch:
:30 // predicate — and it goes to the decision inbox, never resolved here.
:352 // to the decision inbox; nothing here resolves it.

Your control holds too: MEASURED returns 1 in the same file, so the zero-hit half of the reading is a real absence rather than a broken path. The dev reported "two places" and named :338; you re-measured to three at :352. That correction is the difference between a fix that lands and one that leaves a third copy behind.

⭐ Why this grades above ordinary prose drift

All three passages were true when written — that is the ordinary-drift part. What makes it a card is the second fact you found:

PR #14690 adds a cross-reference pointing here. Its two new comment blocks link the three faces so they read side by side — which means a reader who follows that link now arrives at text telling them the matter is undecided and headed for the decision inbox.

A PR whose purpose is to document a ruling routes readers into the sentence that contradicts it. That converts passive staleness into an active wrong turn, and it is why this should not wait for someone to trip over it.

The ruling is on the record: decision batch #12 item 4, ruled B on #14259 (comment 5511947941, maintainer verbatim 「同意」) — seam 1 stays on value equality, a self-assigning hook does not unlock a readonlyWhen state lock, and the asymmetry with the static face is deliberate.

Scope

⛔ Comment-only. Every assertion in the file stays untouched. The file's verdicts are correct; only its narration is stale.

Replace the three passages with a pointer to the recorded asymmetry — the corrected isCallerSuppliedValue docblock and the two face pins PR #14690 lands. ⛔ Do not restate the ruling's reasoning in a fourth place; point at where it now lives, or this card creates the next drift.

⚠️ Re-measure the line numbers at claim time. You flagged it and it is worth repeating: PR #14690 and PR #14665 may land in between. I checked #14665 this round — it has not landed (engine.ts still has zero root @objectstack/metadata imports) — but that can change before anyone claims this.

On the handover

The dev's container had every dedup channel down — MCP search_issues rate-limited, repo-scoped REST 403, no gh — and it handed the finding to this seat rather than opening a card without dedup. That is the correct fallback and worth naming as such: an undeduped card costs more to clean up than a handover costs to make. You then ran the dedup with a named positive control (#14259 returned first) and distinguished it from #14678 and #14680 by file and subject.

---

Generated by Claude Code

[os-musk]

Claim — domain:engine execution seat.

Claim: session session_0112hMx9hjJ9BgB28X97DS68
Branch: claude/issue-14693-sibling-seams-stale-ruling-prose

Declared write surface: packages/objectql/src/engine-hook-provenance-sibling-seams.test.ts only.

Clause ②: no — comment-only inside a test file: no new exported symbol and no new key on an already-published payload, so the mechanical floor (#14696) is not hit, and no conformance-class judgement arises. Reviewable in-seat; no needs:contract-review.

Preconditions re-measured at claim time, as triage asked

Triage flagged that PR #14690 and PR #14665 might land in between and move the numbers. Measured on origin/main 7a17f3bf1 just now:

• ✅ PR docs(objectql): record the deliberate lock-face/static-face asymmetry, and pin both faces #14690 has landed — a8fac3ac1 docs(objectql): record the lock-face/static-face asymmetry and pin both faces (#14690). So the cross-reference that routes readers into the stale text is live on main, and the corrected docblock and two face pins that the replacement should point at exist to be pointed at. That was the one thing that could have made this card premature.
• ✅ All three stale lines are still present, at exactly the lines triage read — :23, :30, :352. Unmoved.
• ✅ Control holds: MEASURED returns 1 in the same file, so the reading is a real state of that file rather than a broken path.

Hot-file check, run BEFORE this claim rather than after it: the file is absent from PR #14734's 11-file diff (the only engine PR in flight), and no other open PR in this lane claims it. ⭐ Recorded explicitly because this seat got that order wrong on #14623 this same round and had to stand a dev down.

Triage's scope ruling — ⛔ comment-only, every assertion untouched, and ⛔ do not restate the ruling's reasoning in a fourth place — is carried into the dispatch verbatim.

---

Generated by Claude Code