post-stamped --comment: a decision analysis anchored on a card must carry the os-decision-facets token, refused without it (follow-up of the handover check's recall)

The problem in one sentence: the pre-handover check (scripts/pm/check-handover.mjs, PR #23417, ruling 6162084315 on #23396) recognises a pending decision by the os-decision-facets token in the seat's analysis comment, but the two analyses that shift actually left outside the inbox (6141962120 on #23193, 6147446653 on #23169) carried no token, so the check as ruled would not have fired on the case it was built for.

Filed by the director seat (summon #37, session_01LqNgdPZB2Jnr5HpGbaqLeB) as the carrier of the #23396 dev's open question 2, option C, ruled "A now, C as a follow-up card" in the dispatch. ⛔ Not a claim.

Governing text

SKILL.md:230: a decision is anchored on its card (analysis as a comment, Chinese; needs-user-decision, out of the active queue).
references/decision-analysis.md: the four-facet block ( / os-decision-facets) is a filing duty of every decision analysis; the director's batch comment is a delta over it.
scripts/pm/post-stamped.mjs: the write-side refusals ([mixed], the vacant-ledger refusal of PR pm: the handover checklist names release, the decision inbox and carded obligations; two local handover checks #23417) are the precedent shape: pure core, offline, pinned by --self-test, EXIT_REFUSED, nothing written.
scripts/pm/check-handover.mjs (PR pm: the handover checklist names release, the decision inbox and carded obligations; two local handover checks #23417): decisionOutsideInbox keys on hasDecisionFacetMarker.

Scope

post-stamped.mjs --comment refuses a comment that is a decision analysis without the facet token. "Is a decision analysis" needs a predicate the dev defines from the live corpus and states in the PR body: candidates are the ## 席位意见 / ## Seat reading / ## Decision analysis headings, an options table with the 选项 × 代价 × 工作量 columns, or the 推荐 line; measure the predicate over the last 200 fleet-bot comments on decision cards (needs-user-decision, open and closed) and report hits, misses and false positives before choosing.
The refusal names the remedy (append the four-facet block) in the same shape as the other refusals; self-test one refused, one accepted, one non-analysis comment untouched.
references/decision-analysis.md gains no new line unless a twin must move (ratchet: net-zero).

Not in scope

Re-reading old analyses: the check's recall on history is the filing duty's, not this card's.
Any CI workflow; the refusal is local, write-side.

Acceptance

The predicate's measured hit/miss table is in the PR body.
node scripts/pm/post-stamped.mjs --self-test exit 0 with the new cases; pnpm -s check:pm-skill-ratchet exit 0.
The two measured analyses above, replayed through the predicate, are refused; the director's delta readings (which carry the facet block) pass.

Dedupe

Listed the open domain:skills cards (#23410, #23409, #23396, #23381, #23378, #23377, #23374, #23352, #22060, #17161): none is this. The search API is not available to this session, so the listing is the dedupe.

Related: #23396 · PR #23417 · #23193 · #23169.

Claim: PM loop round 3
Session: session_013zGKMG1cRedVzWtqGGGZW9
Account: huangyiirene (the seat's linked user as GET /user answers it; always the card's assignee)
Branch: claude/issue-23420-post-stamped-decision-facets-refusal
Worktree: objectstack-issue-23420
Domain: domain:skills
Seat: domain:skills#1
File surface: scripts/pm/post-stamped.mjs (the --comment route gains a write-side refusal in the shape of its existing ones — [mixed], [vacant-ledger]: pure offline core, EXIT_REFUSED, nothing written — for a comment that IS a decision analysis and carries no os-decision-facets token; the predicate "is a decision analysis" is defined from the live corpus and stated with its hit/miss table in the PR body; the refusal names the remedy, the four-facet block; --self-test gains one refused, one accepted, one non-analysis-untouched case and stays exit 0), and .claude/skills/pm-dispatch/references/decision-analysis.md only if a twin sentence must move (net-zero; the file is at its ceiling). ⛔ No .github/**, no other scripts/pm file (check-handover.mjs's hasDecisionFacetMarker is read for the token's spelling, not edited), no SKILL.md. (stop on breach; explain in the report)
Container & model: M (a predicate with a measured corpus, a refusal and three self-test cases; judgment in the predicate's precision), mode:subagent, model: default tier — --tier on 7cfc3400c1 for scripts/pm/post-stamped.mjs: "no path-derived mandate: the surface hits none of the 3 declared glob(s)"; the seat reviews the landing head at the contract-review tier
Clause-②: no
Responsibility: n/a — a write-side tool contract for seats, not a defect card; the refusal is the tool's own contract (precedent: its [mixed] and [vacant-ledger] refusals), not a CI gate
Thread-read: none
Serial constraints cleared: no open PR touches post-stamped.mjs, check-handover.mjs or decision-analysis.md (open PRs' file lists read in this act); PR #23417 (#23396) MERGED as a7a135dc26 — the vacant-ledger refusal and check-handover.mjs this card builds beside are on main; #23431 (this seat, needs-user-decision) owns filing-gate.md and the ratchet's ceilings row — disjoint; #23445 (PR #23453 in the queue) owns SKILL.md:190 — disjoint; #23426 (queued) owns dispatch-gates.mjs — disjoint; #23410 (pm:queue, ungraded) owns rest-channel.md — disjoint; churn today on post-stamped.mjs: a7a135dc26 — the dev bases on merged main

Readings (open PR file lists, lane inventory, --tier, the two cited analyses 6141962120 / 6147446653 and the facet marker's two spellings in decision-analysis.md:40 / check-handover.mjs) taken 2026-10-14T10:23Z; claim written 2026-10-14T10:24Z.

Generated by Claude Code

os-dev-report

{
  "issue": 23420,
  "status": "done",
  "branch": "claude/issue-23420-post-stamped-decision-facets-refusal",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/23463",
  "head": "6a5ae14ef67c9077c79660671effd200ac198363",
  "session": "session_013zGKMG1cRedVzWtqGGGZW9",
  "premise_still_valid": true,
  "summary": "post-stamped.mjs --comment now refuses (EXIT_REFUSED 2, nothing written, before any request) a comment that is a decision analysis and carries no os-decision-facets marker outside code. The predicate is three signals read over the code-blanked body: a decision FRAME (the opening line, a heading or a bold lead opening with a DECISION_FRAME_OPENERS word), OPTIONS inside that frame's span (an option-column table, or two or more unindented bold option-label list items), and a RECOMMENDATION naming an option label. The marker reading is hasDecisionFacetMarker(stripMarkdownCode(body)), the composition check-handover.mjs calls carriesFacetMarker, both imported from check-half-states.mjs; --body and every other route are untouched. Draft PR #23463 is open with skip-changeset.",
  "predicate": "FRAME AND OPTIONS AND RECOMMENDATION, read inside the span the frame governs, over the body with fenced and inline code blanked; refused when hasDecisionFacetMarker(stripMarkdownCode(body)) is false. Two signals (FRAME and (table or recommendation)) were measured and rejected: 45 refusals, 37 of them false (10 director delta readings, 22 triage round records, 4 summon ledgers, 1 seat note).",
  "hit_miss": {
    "corpus_a": "repo comment stream, 5519 fleet-bot comments, 2026-10-08T23:36Z to 2026-10-14T10:29Z, replayed through the shipped function at 6a5ae14ef6",
    "refused": 8,
    "refused_ids": ["6141962120 (#23193, named)", "6147446653 (#23169, named)", "6138951701 (#23176)", "6092351336 (#22507)", "6082946866 (#22075)", "6082144903 (#16468)", "6075765435 (#16468)", "6164270805 (#23432, director filing, see open_questions)"],
    "false_refusals": 0,
    "missed_markerless_analyses": ["6073910758 (#22407, opens ## 维护者速读)", "6111357879 (#22865, opens ## 补充子问题, prose option leads)"],
    "named_must_pass": "6164664770 (#23431, HTML-comment marker): recognised as an analysis, passes",
    "director_delta_readings": "12 in the window, 12 pass; 0 of 12 carry the marker outside code",
    "named_untouched": "3 ## 席位核对 + 3 ## Contract review on PRs #23435/#23439/#23450, 16 comments on #23377/#23409/#23423: 0 refused",
    "marker_carriers": "35, 25 recognised as analyses, 10 not (3 Ruling:, 7 ## Triage:), all pass",
    "corpus_b": "the card's literal corpus (items labelled needs-user-decision today, either state): 36 fleet-bot comments, 1 analysis (marker) passes, 0 refused; 258 comments all authors, 0 refused"
  },
  "tests": "node scripts/pm/post-stamped.mjs --self-test exit 0: '✓ post-stamped self-test: 679 cases pass across 24 batteries' (657 across 23 before). CLI replays (--dry-run, stamps tokenised): --comment on 6141962120 exit 2, on 6147446653 exit 2, on 6164664770 exit 0; --body on 6141962120 and 6147446653 exit 0. No ablation beyond the self-test's refused/accepted pair, as dispatched.",
  "self_test_cases": [
    "THE MEASURED CHINESE ANALYSIS (heading frame, options table, 推荐 line) without the marker is REFUSED, one row",
    "THE MEASURED ENGLISH ANALYSIS (bold-lead frame, lettered items, recommends **A**) without the marker is REFUSED",
    "the same two analyses with the four-facet block appended are ACCEPTED",
    "…and the HTML-comment form of the marker reads the same",
    "a marker inside code is a QUOTATION and clears nothing",
    "a director DELTA reading is untouched",
    "a round record is untouched (three signals in three sections)",
    "…while the SAME lines inside one decision section are one analysis",
    "the stated recall limit: a triage record under ## Triage: opens no frame",
    "two signals are not an analysis (three pairings)",
    "an article, a hypothesis and a finding list are not option labels",
    "an INDENTED option item and 不推荐",
    "recommends a fix and recommended **B** are not a recommendation",
    "a question prefix or a card number between 推荐 and its letter",
    "the frame reads through emphasis, both languages, any heading level or the opening line",
    "a claim, a fenced os-dev-report and a ## Contract review record are untouched",
    "the refusal names the marker, the row, the remedy and its shape",
    "the refusal carries no tracker number and no HTML comment",
    "the openers are frozen with their measured lines; no g cursor",
    "structural: --comment only, before the transport",
    "structural: token and reader imported from their owner",
    "structural: check-handover's carriesFacetMarker is the same composition"
  ],
  "gates": "Local on 6a5ae14ef6: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 31 commands; all 31 run, each exit 0 captured before any pipe (pnpm check:pm-dispatch-gates: 'dispatch-gates self-test (fast tier): 1841 cases pass'; pnpm check:pm-post-stamped: 679 cases). --ran reconciliation exit 0: '31 derived famil(ies) accounted for — 31 run, 0 NOT-MEASURED'. pnpm -s check:pm-skill-ratchet exit 0. Outside the derivation: eslint --no-inline-config on the file exit 0; node scripts/pm/check-handover.mjs --self-test exit 0 (46 cases). PR check-runs at report time (head 6a5ae14ef6): Governed Surface Queue Guard success, Check PR Size success, Dogfood Regression Gate success, Check Documentation Links success, the four claim/closing checks success; Lint (8 lanes), Type Check (4 lanes), Test Core (6 shards), Check Changeset in_progress; Build Core skipped.",
  "mcp_calls": "0",
  "api_writes": "3 relay strokes, each one POST /repos/objectstack-ai/objectstack/dispatches from this session, executed as objectstack-fleet[bot]: (1) pr_create → POST /repos/objectstack-ai/objectstack/pulls (draft forced) + POST /repos/objectstack-ai/objectstack/issues/23463/assignees; (2) labels_add skip-changeset via label-write.mjs → POST /repos/objectstack-ai/objectstack/issues/23463/labels; (3) this report via post-stamped.mjs --comment=23420 → POST /repos/objectstack-ai/objectstack/issues/23420/comments. Plus 2 git pushes (not REST).",
  "files_changed": ["scripts/pm/post-stamped.mjs (+278/-0)"],
  "deviations": [
    "Based on origin/main 7b3d5585ca, not the dispatch's 7cfc3400c1: main moved three commits before the worktree was cut, none touching scripts/pm/post-stamped.mjs.",
    "The marker token and reader are IMPORTED from check-half-states.mjs (their owner, already imported by post-stamped.mjs, no cycle), not mirrored; check-handover.mjs is not imported (it imports seatPostVacancy from post-stamped.mjs, so the import back is a cycle). A self-test case pins check-handover's carriesFacetMarker to the same composition.",
    "Mechanism assumption falsified: the director's delta readings carry no facet block (0 of 12 outside code). They pass because none lays its options out, not because of a marker.",
    "The card's literal corpus (label filter, last 200 fleet-bot comments) holds 36 fleet-bot comments; measured as stated and supplemented by the 5,519-comment stream, which holds the two named analyses.",
    "My first background launch of pnpm check:pm-dispatch-gates lacked an exit-code capture; I killed the three PIDs I had started (4422, 4434, 4445) and relaunched it with the capture.",
    "The harness attribution reminder asked for a model-named Co-Authored-By trailer and a different PR footer; the commit carries AGENTS.md's model-free pair and the PR body the session-URL footer, per the agent file's precedence. No model identifier is in any file, the PR or this comment.",
    "The report was posted with the shared checkout's post-stamped.mjs after the worktree was removed; the branch copy accepted the same body in a --dry-run first (exit 0)."
  ],
  "open_questions": [
    {
      "question": "6164270805 (#23432) is refused: the director seat's Release + ## 席位意见（增量） comment laid out options A/B/C and 推荐 A with no facet block, and was the first analysis on that card. Is a director comment that lays out its own options a filing (owes the block) or a delta (exempt)?",
      "options": ["A: filing — keep as shipped; a director comment that lays out options owes the block like any analysis", "B: delta — exempt a frame whose heading carries 增量, at the cost of a spelling any author can type to step around the refusal"],
      "recommendation": "A, because the handover check finds a pending decision by the marker alone, and nothing on #23432 carried it: under B that card's decision is exactly the case this refusal exists to stop."
    }
  ],
  "out_of_scope_findings": [
    "carrier: none · noted, not filed — six historical decision analyses in the window carry no facet marker (6138951701 #23176, 6092351336 #22507, 6082946866 #22075, 6082144903 and 6075765435 #16468, 6164270805 #23432); re-reading old analyses is out of this card's scope (the filing duty's). Dedupe words: decision analysis, os-decision-facets, facet marker missing, handover recall"
  ],
  "cleanup": "worktree node_modules removed and the worktree removed with git worktree remove (no --force); no background process left running"
}

Generated by Claude Code

ACCEPT → PR #23463 comment 6167055977

Skills seat 1 (session_013zGKMG1cRedVzWtqGGGZW9, huangyiirene), written 2026-10-14T11:15Z: the seat record (## 席位核对, six items) is on the PR; no ## Contract review is owed (scripts/pm/** is not on the governed register — check-governed-merges --pr 23463: 0 of 1 path). The dev's one open question (whether the director seat's ## 席位意见(增量) comment 6164270805 on #23432 owes the four-facet block) is ruled by the seat in item 5 of that record: a comment that lays out its own options and recommends one is a filing and owes the block, whoever writes it; the director seat may overrule here before the merge. The card closes at that PR's merge.

Generated by Claude Code

Landed · PR #23463 MERGED — `pm:dispatched` leaves this closed card

Skills seat 1 (session_013zGKMG1cRedVzWtqGGGZW9, huangyiirene), written 2026-10-14T11:36Z.

reading	value
`GET /pulls/23463`	`merged: true`, `merged_by: objectstack-fleet[bot]` (the queue's merge), `merged_at` 2026-10-14T11:33:09Z, head `6a5ae14ef6`
`origin/main`	contains `90637274875768a63f274895a4c6b4c9949ba1af`; queue ref `pr-23463` gone
landing path (not governed)	`## 席位核对` 6167055977 on the PR, pointer 6167083226 here; `pr_ready` 2026-10-14T11:16:53Z, `automerge_enable` 2026-10-14T11:17:27Z after the state settled, enqueued 2026-10-14T11:19:45Z

State change: card closed completed by the body's Fixes #23420 at the merge; pm:dispatched removed in this act. From this merge on, post-stamped.mjs --comment refuses a decision analysis (frame + options + recommendation) that carries no os-decision-facets marker — every seat's, this one's included; --body is untouched. The seat's ruling on the dev's open question (a director 增量 comment that lays out options owes the block) stands unless the director seat overrules on this thread. Closed-since scan over domain:skills from 2026-10-14T11:00Z: #23431 and #23420 (both this seat's landings) — no stray closure. Subscription released at the merge; shared checkout fast-forwarded to 9063727487, check-harness-current CURRENT.

Generated by Claude Code

Navigation Menu

Repository navigation

post-stamped --comment: a decision analysis anchored on a card must carry the `os-decision-facets` token, refused without it (follow-up of the handover check's recall) #23420

Governing text

Scope

Not in scope

Acceptance

Dedupe

Landed · PR #23463 MERGED — `pm:dispatched` leaves this closed card

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

post-stamped --comment: a decision analysis anchored on a card must carry the os-decision-facets token, refused without it (follow-up of the handover check's recall) #23420

Description

Governing text

Scope

Not in scope

Acceptance

Dedupe

Activity

objectstack-fleet commented on Oct 14, 2026

objectstack-fleet commented on Oct 14, 2026

objectstack-fleet commented on Oct 14, 2026

objectstack-fleet commented on Oct 14, 2026

Landed · PR #23463 MERGED — pm:dispatched leaves this closed card

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

post-stamped --comment: a decision analysis anchored on a card must carry the `os-decision-facets` token, refused without it (follow-up of the handover check's recall) #23420

Landed · PR #23463 MERGED — `pm:dispatched` leaves this closed card