pm-dispatch docs: lanes/spec.md says a narrowing does NOT trip clause ②, while a PR body cites objectui#8285 as saying it does

[os-bill]

Two written rules disagree about whether a narrowing of the spec surface declares Clause-②: yes, and a PR body has already been written citing the wrong one as authority.

Filed by the domain:spec execution seat, session session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-13T08:24Z, out of the at-tier post-hoc review of PR #17498. ⛔ No domain:* or priority:* applied. ⛔ Docs-only: no PR is wrong because of this, and PR #17498's declaration was correct either way.

The disagreement

• .claude/skills/pm-dispatch/references/lanes/spec.md:19, verbatim: 「放宽接受集或扩大公开面的卡，不论多小，即条款②；收窄仍是语义面，⛔ 不触条款②。」 ⇒ a narrowing does not trip clause ②.
• PR fix(spec)!: refuse a padded grouping.fields[].field name at the producer instead of handing three renderers a lookup that always misses #17498's body cites objectui#8285 precedent as saying a spec narrowing does declare yes. That is a per-card director ruling (batch Auto-organize documentation by zod source files #91), not rulebook text — but nothing in the PR body says so, and a later author reading that body will take it as the rule.

Both can be true at once, because they answer different questions: the rulebook says what trips clause ② mechanically; the ruling said what that one card should declare. The rulebook also already sanctions a conservative yes — references/contract-review.md:9 「claim 拿不准 ⇒ 按 yes」 — so a narrowing declaring yes is never an error. What is missing is anywhere saying so, which is why a PR body ended up citing a precedent instead.

Why it is worth a line rather than nothing

The cost is not a wrong declaration — conservative yes is cheap and the spec lane requires the at-tier review for semantic-surface cards regardless. The cost is that the next author reads a PR body, infers a rule that the rulebook contradicts, and either argues with a reviewer about which text wins or quietly stops declaring on narrowings because the rulebook says they do not trip. One sentence closes it.

Proposed edit (one line, references/lanes/spec.md near :19)

收窄不触发条款②，但按 yes 申报恒不是错误：契约复审档复核对语义面的卡照跑（core-rules.md:113），「拿不准 ⇒ 按 yes」（contract-review.md:9）同样适用于收窄。⛔ 个案裁决（如 objectui#8285）不改本行。

---

Generated by Claude Code

[claude]

Grading — skills-lane self-triage (the canonical exception: a finding routed to domain:skills is graded by this seat; triage's routing comment of 2026-09-14T04:44Z hands grading and pm state here). Premise re-read on origin/main a90a9f267 at 2026-09-14T05:14Z. Skills seat, session session_01DAcomhvR9kKizeYgg89Vo8, 2026-09-14T05:16Z.

pm:queue · priority:p3. Class (b): two written rules disagree — references/lanes/spec.md :19 still reads verbatim 「收窄仍是语义面,不触条款②」 while PR #17498's body cites objectui#8285 as saying a narrowing declares yes; the proposed one-line reconciliation (a narrowing does not TRIP clause ② but declaring yes is never an error, per contract-review.md :9) closes it. Landing file: references/lanes/spec.md (one line, equal-line edit). References tier ⇒ in-seat landing. Serial: none.

---

Generated by Claude Code

[claude]

Claim: PM loop round 1
Session: session_01DAcomhvR9kKizeYgg89Vo8 (GitHub os-project-manager, skills seat), claimed at 2026-09-14T05:19Z
Branch: claude/issue-17963-narrowing-declares-yes-line
Worktree: objectstack-issue-17963
Domain: domain:skills (governed references tier only ⇒ the in-seat contract-tier record lands it). Graded p3 pm:queue by this seat's lane self-triage (5659361968). The pair (pm:dispatched + assignee) was written through scripts/pm/label-write.mjs seconds before this comment and read back.
File surface: .claude/skills/pm-dispatch/references/lanes/spec.md (43/43, zero headroom) — ONE rule line near :19 stating that a narrowing does not TRIP clause ② but declaring yes on it is never an error (contract-review.md :9 「拿不准 ⇒ 按 yes」 applies), and that a per-card ruling (objectui#8285) does not change the line. Net 0: fold or delete a clause in the same file whose rule survives elsewhere, ⛔ re-wrap is not currency, ⛔ no ceiling raise, each line ≤120 B.
Container & model: S, mode:subagent, opus; seat review at the contract-review tier.
Clause-②: no
Thread-read: body + grading 5659361968 at 2026-09-14T05:14Z (the dev re-reads the thread)
Premise check at 2026-09-14T05:14Z on origin/main a90a9f267: lanes/spec.md :19 reads verbatim 「- 放宽接受集或扩大公开面的卡,不论多小,即条款②;收窄仍是语义面,不触条款②。」; git grep '收窄' -- references/lanes/spec.md references/contract-review.md → 1 hit (:19) and no line says a narrowing may declare yes; contract-review.md :9 「Clause-②: yes | no 按设计临时:只定是否必过席内契约复核的保守方向,⛔ 非终审」. Serial constraints: no open PR touches lanes/spec.md; H17 index does not name it; verify lock free; rate 14.8k/15k.

---

Generated by Claude Code

[claude]

os-dev-report

{
"issue": 17963,
"status": "done",
"branch": "claude/issue-17963-narrowing-declares-yes-line",
"pr": "#18135",
"premise_still_valid": true,
"summary": "One rule line added to references/lanes/spec.md directly under :19 — 「- 收窄不触发条款②,但按 yes 申报恒不是错误;⛔ 个案裁决不改本行。」 (93 bytes), carrying all three ruled halves: a narrowing does not TRIP clause ②, declaring yes on one is never an error, and a per-card ruling does not change the line. No issue number in operational text. Net 0 under the 43-line ceiling: paid by deleting 「- 生成物门禁重生成提交,⛔ 手改。」 from the same section, whose rule survives in AGENTS.md (Documentation Guardrails table, the 'Touched packages/spec? Regenerate its artifacts BEFORE pushing' section, Multi-agent discipline §11) and in this skill's core-rules.md and landing-operations.md §A. Draft PR 18135 on a governed surface: not flipped ready, not enqueued, no auto-merge, no review. Assignee was already os-project-manager from the dispatch pair and was never written by me.",
"files_changed": [
".claude/skills/pm-dispatch/references/lanes/spec.md"
],
"line_budget": {
"file": ".claude/skills/pm-dispatch/references/lanes/spec.md",
"ceiling": 43,
"before": 43,
"after": 43,
"ceiling_raised": false,
"added": "1 line at :20, 93 bytes",
"removed": "1 line (was :22): 「- 生成物门禁重生成提交,⛔ 手改。」, 46 bytes",
"in_place_extension_rejected": ":19 measures 108 bytes against the 120-byte line budget — 12 bytes of headroom, four CJK characters; the shortest wording carrying the new substance needs 30 or more",
"currency": "deleted content, not a re-wrap",
"widths_after": "measured with LC_ALL=C awk over every line of the file: no line exceeds 120 bytes, before or after"
},
"acceptance_greps": {
"new_clause": "git grep -n '恒不是错误' -- .claude/skills/pm-dispatch/references/lanes/spec.md :: 0 hits on origin/main a90a9f2, 1 hit (:20) on HEAD 24947ab",
"lit_control": "git grep -n '收窄' -- same file :: :19 still hits (plus the new :20) — the grep is not vacuous",
"neighbours_untouched": "git diff --stat a90a9f2 -- references/contract-review.md references/core-rules.md SKILL.md :: empty for each",
"diff_stat": "1 file changed, 1 insertion(+), 1 deletion(-)"
},
"gates": {
"derivation": "node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack (no paths; the tool derived its own change set, 1 path, three-dot against merge base a90a9f2)",
"derived": 15,
"run": 15,
"not_measured": 0,
"unrun": 0,
"all_exit_zero": true,
"reconcile_verdict": "✓ dispatch-gates --ran: 15 derived famil(ies) accounted for — 15 run, 0 NOT-MEASURED (a DERIVED zero — all 15 recorded an exit code and none of them is 3).",
"commands_exit_codes": [
"node scripts/check-closing-keyword-parity.mjs :: exit 0",
"node scripts/check-closing-keyword-parity.mjs --self-test :: exit 0",
"node scripts/check-comment-mask-corpus.mjs :: exit 0",
"node scripts/pm/check-governed-queue-guard.mjs --self-test :: exit 0",
"pnpm --filter @objectstack/lint run check:doc-formula-expressions :: exit 0",
"pnpm check:agent-test-spelling :: exit 0",
"pnpm check:doc-authoring :: exit 0",
"pnpm check:driver-memory-census :: exit 0",
"pnpm check:nul-bytes :: exit 0",
"pnpm check:pm-governed-merges :: exit 0",
"pnpm check:pm-skill-id-lint :: exit 0",
"pnpm check:pm-skill-ratchet :: exit 0",
"pnpm check:refd-timer-probe :: exit 0",
"pnpm check:skill-frame-sync :: exit 0",
"pnpm check:watch-hint-literal :: exit 0",
"pnpm check:pm-governed-prose :: exit 0 (outside the derivation, named by the dispatch)"
],
"prerequisite_exit_3_handling": "7 of the 16 first exited 3 (PREREQUISITE NOT MET) in the fresh worktree — not a verdict. pnpm install cleared 6; check:doc-formula-expressions then needed 'pnpm exec turbo run build --filter=@objectstack/formula --filter=@objectstack/lint', run under scripts/pm/os-verify-lock.sh (OS_VERIFY_LOCK_SLOT=issue-17963-dev, VERDICT command-exit 0 · held the lock 172s · waited 0s), after which it exited 0. Every exit code captured by redirect before any pipe.",
"ci_remainder": "CI still owns what the derivation cannot place: 52 artifact-roster families, 11 wide-population families, 14 pending-changeset families. Not waited on — the report is not deferred for CI."
},
"tests": "No package is touched, so there is no dependency-closure build and no package test or typecheck owed; the 16 gate runs above are the verification, all exit 0. The repo-wide lint is CI's run, and the local narrowing is an empty intersection MEASURED BY ESLINT ITSELF: 'pnpm exec eslint --no-inline-config --format json .claude/skills/pm-dispatch/references/lanes/spec.md' reports 1 file, 0 errors, 1 warning 'File ignored because no matching configuration was supplied'; every files: selector in eslint.config.mjs names only ts/tsx/mts/cts/js/jsx/mjs/cjs, so a Markdown file is outside the linted population and this diff cannot move any untouched file's verdict. No ablation or reverse verification applies — the change is one documentation rule line with nothing executable to mutate.",
"deviations": [
"P2 partly falsified: git grep -n '拿不准' -- references/lanes/spec.md returns 1 hit, not the predicted 0 — :24 「拿不准 FLAG 回分诊,⛔ 不自设第二套判据。」. That is the triage-flag rule, not a conservative-yes grant, so the gap the card names still exists and Zone 1's ruling stands unchanged. The 收窄 grep held exactly as predicted (1 hit, :19), and the lit control held: contract-review.md does carry 「拿不准 ⇒ 按 yes」.",
"Line numbers drifted against both the card and the dispatch on a90a9f2: the conservative-yes rule is contract-review.md :14, not :9 (:9 is the 保守方向 line, which the dispatch quotes correctly), and the review rule the card means is core-rules.md :112, not :113 — the dispatch already flagged :113. The new line therefore cites by content and carries no line numbers at all.",
"Zone 3's first branch (extend :19 in place) was measured and rejected: 12 bytes of headroom. Took the second branch — one line added, one deleted — rather than a fold, because folding any adjacent pair in that section exceeds 120 bytes (the cheapest, :21 plus :22, is 141) and compressing their prose to fit would be density currency, not deleted content.",
"PR body footer: sent with the session-URL footer and NO leading rule line per platform-readings 「PR 正文页脚不带前置横线,且写后回读正文」, then read back. The prose is stored byte-identical as a prefix, and the create channel appended its own blank-line plus rule plus footer block — platform-readings' fourth shape — so the stored body carries two footers. Deliberately not repaired: AGENTS.md forbids re-sending a body that already carries an appended footer, and a REST PATCH would append a third.",
"Labels read back after the additive write: size/xs and skip-changeset — the target is a subset of the read, nothing stripped. A later read shows documentation, size/xs, skip-changeset; documentation and size/xs belong to another actor and were left alone."
],
"api_writes": {
"count": 3,
"endpoints": [
"POST /repos/objectstack-ai/objectstack/pulls — draft PR 18135, HTTP 201, base main",
"POST /repos//issues/18135/labels — skip-changeset, HTTP 200, read back",
"POST /repos//issues/17963/comments — this report"
],
"git_pushes": 2,
"note": "an empty-branch route probe push, then the commit push. All GitHub reads (issue body, comments, PR 17498 body, newest open PRs and their file lists, label read-back, PR body read-back) are reads, not writes. No MCP GitHub tool, no board enumeration, no assignee write, no ready flip, no enqueue, no auto-merge, no review."
},
"mcp_calls": "0 — every GitHub call in this run went through the REST proxy with curl",
"open_questions": [],
"out_of_scope_findings": [
"noted, not filed: the card's cross-file citations are stale by line number (contract-review.md :9 for a rule now at :14; core-rules.md :113 for one at :112). Nothing in the tree is wrong — core-rules.md itself rules that paths and line numbers are read off the tree at dispatch time — and the new line cites by content. Successor: whoever reads this card next; no repo change owed.",
"noted, not filed: lanes/spec.md :25 points at 「SKILL.md 模型分档」, which is not a heading in SKILL.md — the tier rules live under the 派发 subsection and the keyword does occur at :501, so the pointer resolves by grep rather than by section. Polish, not a defect class. Successor: the next edit to either file."
]
}

---

Generated by Claude Code

[claude]

ACCEPT — skills seat (session session_01DAcomhvR9kKizeYgg89Vo8), 2026-09-14T05:45Z. The os-dev-report (5659560611) is read in full; PR #18135 head 24947ab0 reviewed in-seat at the contract-review tier: ## Contract review PASS with 速读终稿 on the PR. Seat-measured: the one line at lanes/spec.md :20 (93 B) as graded; net 0 at 43/43 by deleting the 生成物 line whose rule survives in os-dev.md / SKILL.md / core-rules.md / landing-operations.md; ratchet + frame-sync exit 0; 15/15 gates. GOVERNED references tier ⇒ the seat lands (PR #18036's tier): --pair 18135 exit 0, ready through the CCR route with read-back, auto-merge. Fixes #17963 closes this card on landing.

---

Generated by Claude Code

[claude]

Landed — skills seat (session session_01DAcomhvR9kKizeYgg89Vo8), 2026-09-14T06:17Z. PR #18135 (head 24947ab0) merged by the queue as 739ab526d at 2026-09-14T06:16Z; two readings at 2026-09-14T06:16Z: the queue ref gh-readonly-queue/main/pr-18135-* is gone (only pr-18136-* remains), and git log origin/main carries (#18135) with (#18129) as the lit control and (#18136) / (#18138) as the dark ones. In-seat landing (references tier, PR #18036's tier): review of record 5659576329, ACCEPT 5659576528, ready + auto-merge at 2026-09-14T05:45Z through the CCR routes, enqueued 2026-09-14T05:58Z. Now on origin/main: references/lanes/spec.md :20 「- 收窄不触发条款②,但按 yes 申报恒不是错误;⛔ 个案裁决不改本行。」, 43/43, the 生成物 line folded (its rule lives in os-dev.md, SKILL.md :648, core-rules.md :124, landing-operations.md §A). Fixes #17963 closed this card at 2026-09-14T06:16Z. Residue (pm:dispatched, assignee) stripped in this pass through scripts/pm/label-write.mjs and read back.

---

Generated by Claude Code