You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
os-half-state-sweep — machine-findable marker for this generated view.
Generated view — not a second tracker. Authority lives on each card and PR (one-board rule); this body is rewritten IN PLACE by the scheduled patrol workflow (.github/workflows/half-state-patrol.yml) on every run, and the edit history is the archive. Report-only: every row is patrol input, never a gate verdict, and this sweep never fixes a state. Each predicate and the protocol clause it enforces are documented in scripts/pm/check-half-states.mjs.
The timestamp above is the patrol's own heartbeat: a Swept line that stops advancing means the standing caller died, which is the failure this anchor was created to make visible. Read it before you read the rows.
⚠️2 UNJUDGED row(s) in this sweep — an input this patrol could NOT read, not a state it read and found clean. They are sorted above the ordinary rows so the body's size trim can never be what removes them (#4690, #11218), and they are the rows to judge BY HAND: nothing in a later sweep will resolve them on its own.
check-half-states: swept 281 open pm-/p0-labeled issue(s), 356 open issue(s) in the unscoped pass (H13–H15, H18), 25 open PR(s) (merge state read on 0 of 0 H16 candidate(s)) and 397 recently-merged PR(s) in objectstack-ai/objectstack — 262 half-state(s) found. H22 read 400 recently-closed issue(s) for pm:* state residue (bounded window; older closed carriers are outside it by design). H23 read 300 squash commit message(s) from the default branch's recent window, carrying 3 closing-keyword binding(s) across 3 message(s) (bounded window; a message that landed before it is invisible by design). Hold comments read on 78 of 78 H17 candidate(s). Blocked-by: comment fallback read on 43 of 43 candidate(s). Restart-when: hold comments read on 33 of 33 H9 candidate(s). Blocker liveness (H19): targets resolved on 23 of 25 distinct Blocked-by: target(s) named by open pm:blocked card(s) — the 2 unresolved target(s) are named on their own cards' rows, and those rows sort ABOVE the size trim so they cannot be what a truncated body drops (#11218: this clause used to be an unconditional promise, and on the 2026-08-25T02:08Z sweep it was false — 199 rows were trimmed and not one rendered row carried an unresolved target) Cross-repo reachability was measured directly on 1 sibling repo(s), of which 1 do(es) not answer this credential — those targets are unjudgeable by ruling (each install reads its own repo with its own repo-scoped token) and ⛔ no re-run resolves them. Dispatch liveness (H20 + H27): remote branch read on 7 of 7 distinct claimed branch(es) named by open pm:dispatched card(s) past the 60-minute threshold — one read serving both rows, so H27's 24h population is a subset of this one and costs no request of its own. Seat liveness (H32): marker thread read on 5 of 5 HELD seat post(s) whose lane is countable on THIS board — a seat held for a sibling repo's lane is out of scope here (its inventory is unreadable from this sweep, so an empty-looking queue would mean nothing), and an unread thread makes H32 decline to judge that seat rather than accuse it. Gate-removal patrol (H35): 0 removal(s) of a gate-semantic label read from 0 page(s) of the repo-wide issue-event stream over the last 12h — no per-card timeline fetch. 0 of them are UNJUDGEABLE (a gate that only ever had ONE carrier leaves 「双载体同笔清标」 no evidence in either direction, so neither H31 nor H35 can say cleared-or-stripped). Shared-file holds (H36): changed-file page read on 24 of 24 open PR(s) — a pair needs both sides read, so a shortfall can only MISS a hold, never invent one. Family folds (H37): 2 live shared branch(es) claimed by more than their own chain head, and a member comment page read on 99 of 99 open pm:queue card(s) — that second read is bought ONLY when a fold is live, so 0 of 0 is a board with no fold in flight rather than a pass that skipped one, and an unread member can only MISS a drifted write, never invent one. Report-only: findings are patrol input, not a gate verdict.
Findings
H19#10938 — pm:blocked and 1 of 1 Blocked-by: target(s) could NOT be resolved this sweep (objectstack-ai/cloud#944 (HTTP 404; objectstack-ai/cloud is NOT readable to this sweep's credential)) — so whether this block has outlived its blocker is UNJUDGED, not confirmed. Unread is not still-open (check:react-declaration-parity 是唯一没接进任何 workflow 的源码审计门禁,且无 MANIFEST 时静默 skip 退出 0 —— 它现在永远不可能红 #4690): a target dropped in silence reads as a healthy block forever, which is the exact failure this item exists to end, so it is named here instead. ⚠️ UNJUDGED is not a quiet row and must not be skimmed as one: this card's block is exactly as unverified as if nothing had been read at all. ⚠️ 1 of them are unjudgeable for a reason that has NOTHING to do with this card: the repo itself does not answer this sweep's credential (measured directly, by a separate GET /repos/<owner>/<name> — not inferred from the issue 404). That is the cross-repo class the contract-first split manufactures, and it is a standing, ACCEPTED limit rather than a defect to chase: each patrol install reads its own repo with its own repo-scoped token by ruling, so no re-run and no re-read of this card will ever resolve these. ⛔ Do not "fix" it on the card — judge the target BY HAND, or take a credential change to routing/security, whose call it is. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H19#11225 — pm:blocked and 1 of 1 Blocked-by: target(s) could NOT be resolved this sweep (objectstack-ai/cloud#1509 (HTTP 404; objectstack-ai/cloud is NOT readable to this sweep's credential)) — so whether this block has outlived its blocker is UNJUDGED, not confirmed. Unread is not still-open (check:react-declaration-parity 是唯一没接进任何 workflow 的源码审计门禁,且无 MANIFEST 时静默 skip 退出 0 —— 它现在永远不可能红 #4690): a target dropped in silence reads as a healthy block forever, which is the exact failure this item exists to end, so it is named here instead. ⚠️ UNJUDGED is not a quiet row and must not be skimmed as one: this card's block is exactly as unverified as if nothing had been read at all. ⚠️ 1 of them are unjudgeable for a reason that has NOTHING to do with this card: the repo itself does not answer this sweep's credential (measured directly, by a separate GET /repos/<owner>/<name> — not inferred from the issue 404). That is the cross-repo class the contract-first split manufactures, and it is a standing, ACCEPTED limit rather than a defect to chase: each patrol install reads its own repo with its own repo-scoped token by ruling, so no re-run and no re-read of this card will ever resolve these. ⛔ Do not "fix" it on the card — judge the target BY HAND, or take a credential change to routing/security, whose call it is. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H11#1883 — important card parked: priority:p2 + security sitting in pm:on-hold, open ~76d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card's Blocked-by: / Restart-when: liveness in the triage round.
H11#3267 — important card parked: priority:p2 sitting in pm:on-hold, open ~42d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card's Blocked-by: / Restart-when: liveness in the triage round.
H5#6015 — title says 🟢 session_011c4YfanSNzNEVaHhDuSAfB but assignees are [none]
H5#6023 — title says 🟢 os-elon but assignees are [os-zhuang]
H30#7251 — pm:queue with ~10d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H22#7369 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H15#7497 — oldest UNCLAIMED pm:blocking card: open ~461h, 4 of 4 open pm:blocking card(s) unassigned. The lane selection order puts pm:blocking second only to priority:p0, so an unclaimed one aging while fresher cards are picked is selection-order drift — visible here by name instead of only in a seat's memory. Age is the CARD's (created_at, the same quantity the order's within-rank tie-break reads), not the label's: that would need a per-card timeline fetch this sweep never makes. Visibility row — no threshold, it reports unconditionally, and like everything here it is patrol input, not a verdict.
H22#8103 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H11#8343 — important card parked: bug sitting in pm:on-hold, open ~17d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card's Blocked-by: / Restart-when: liveness in the triage round.
H9#9139 — pm:on-hold with no Restart-when: line in EITHER channel — not in the body, and not in any comment on the thread (both were read) — the hold state is legal only with a machine-fireable exit (Restart-when: closed <owner/repo>#N, or a one-line executable predicate). ⚠️ READ THE BODY AND THE THREAD BEFORE ACTING: a line that IS there but which this scan cannot parse looks exactly like an absent one. Decoration is tolerated (backticks, a -/* bullet, ** bold), but a mis-spelled or lowercased key is not — the unlock scan greps the literal. If the line is there, unwrap or re-spell it; that is the whole fix, and no state change is due. Add or repair the line first. Closing is the LAST resort and applies only to a card that genuinely has no fireable exit: such a card is closed not planned with reason + provenance in the closing comment (type:Bug holds re-route instead — see the state model's Bug branch). Channel: like Blocked-by: (H4/H14), a Restart-when: line counts from the body OR a comment — either channel discharges the duty, and the unlock scan reads both (hold protocol: widen the Restart-when: channel contract to body OR comment — unlock scan + H9 + the SKILL.md hold-row text move together (a comment-channel exit silently expired for 2 days) #10403 closed the old body-only gap).
H22#9276 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H30#9788 — pm:queue with ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H31#10025 — needs:contract-review on the CARD while its delivering open PR feat(service-automation): definition-level input-schema refusal is non-retryable (FLOW_INPUT_SCHEMA_INVALID) #13371 (draft) does NOT carry it — the two carriers of one gate disagree, so the pair was written half way: either the hang never reached the PR side (「PR 一存在即挂」, and the PR exists), or a PASS cleared the PR side and stopped there, leaving the card gated behind a review that has already passed. The gate is a DUAL carrier — 「两边都挂好」, hung in one stroke and cleared in one stroke, each carrier written read-modify-write with a READ-BACK (「闸门被剥不是红灯是放行」: a stripped gate is a GREEN light, and 「被剥」 and 「从未挂过」 are indistinguishable in the evidence, so the read-back is the only way either is ever noticed). Report-only: ⛔ never a label written from this script — hanging or clearing a review gate from a sweeper would be issuing the verdict, which is 自查放行.
H11#10032 — important card parked: type:Bug sitting in pm:on-hold, open ~11d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card's Blocked-by: / Restart-when: liveness in the triage round.
H30#10374 — pm:queue with ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H5#10602 — title says ⏳ vacant but assignees are [os-warren]
H30#10622 — pm:queue with ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H8#11042 — delivering PR fix(qa,spec,client): address the metadata write door by its canonical singular type segment #13011 (merged 2026-08-29) is MERGED but the card still carries pm:dispatched — the merge's paired write never landed. Drop pm:dispatched and re-grade the remainder (re-queue, close, or block the un-delivered half) in the same stroke, and 「同笔摘 assignee」 — the landing re-label owes the ASSIGNEE DROP too. A re-graded card that keeps the finished dev's assignee lands straight in H24's two-views contradiction (pm:queue + assigned = dispatchable to the queue view, taken to the claim rule), which is how 17 cards across three repos got stuck where nobody could legally move them (finding(pm): open+pm:queue+assignee 双向撒谎态,三仓 17 张——四条机器门修法(新 H 规则/配对写入补 assignee/巡查跨仓/H22 窗口),外加一个需裁决的「等人工动作」状态 #11196). ⚠️ Agent identity only: a HUMAN assignment may be real ownership and is ⛔ never cleared by an agent.
H11#11225 — important card parked: security sitting in pm:blocked, open ~7d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card's Blocked-by: / Restart-when: liveness in the triage round.
H26#11286 — pm:blocked on 1 target(s) that can never CLOSE: #7497 (pm:on-hold). The unlock predicate is "the Blocked-by: target closed", and pm:on-hold / needs-user-decision are by definition states a card sits in WHILE OPEN — so this block has NO MECHANISM THAT WILL EVER RELEASE IT. Every existing check passes on this card (the line is present, the target resolves, the target is open, the label is correct), which is why the measured instances were found by a human reading and by no gauge; H9 asks the mirror question about the HELD card and nothing asked about the WAITING one. ⚠️ Not a claim that the block is wrong — waiting on a deferred card is sometimes exactly right. It says the wait is indefinite BY CONSTRUCTION, so the release has to come from the target's own state changing (a ruling answered, a hold restarted) and someone has to want that.
H9#11453 — pm:on-hold with no Restart-when: line in EITHER channel — not in the body, and not in any comment on the thread (both were read) — the hold state is legal only with a machine-fireable exit (Restart-when: closed <owner/repo>#N, or a one-line executable predicate). ⚠️ READ THE BODY AND THE THREAD BEFORE ACTING: a line that IS there but which this scan cannot parse looks exactly like an absent one. Decoration is tolerated (backticks, a -/* bullet, ** bold), but a mis-spelled or lowercased key is not — the unlock scan greps the literal. If the line is there, unwrap or re-spell it; that is the whole fix, and no state change is due. Add or repair the line first. Closing is the LAST resort and applies only to a card that genuinely has no fireable exit: such a card is closed not planned with reason + provenance in the closing comment (type:Bug holds re-route instead — see the state model's Bug branch). Channel: like Blocked-by: (H4/H14), a Restart-when: line counts from the body OR a comment — either channel discharges the duty, and the unlock scan reads both (hold protocol: widen the Restart-when: channel contract to body OR comment — unlock scan + H9 + the SKILL.md hold-row text move together (a comment-channel exit silently expired for 2 days) #10403 closed the old body-only gap).
H30#11585 — pm:queue with ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H22#11646 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H10#11663 — priority:p0, open and unassigned, with no activity for ~124h (threshold 24h) — p0 is queue-jump priority, so a stale unclaimed one usually means no seat's declared scan scope covers the queue it sits in. Put it in the triage round brief / decision box and name its reader.
H19#11668 — pm:blocked while 1 of 1 Blocked-by: target(s) — read from body OR comment — is CLOSED (#9689 (closed 2026-08-24T14:39:13Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H22#11728 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H30#11742 — pm:queue with ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H4#11809 — pm:blocked with a Blocked-by: line in NEITHER channel — not in the body, and not in any comment on the thread (both were read). Either channel discharges the duty: seats park the line in a comment on purpose, because rewriting a body through the MCP escaping hazard ([finding] GitHub MCP issue reads return HTML-escaped bodies, so any seat that appends a line by rewriting a card body may store ' literals into its code blocks #8813) is the riskier write. So this is not a formatting nit — no machine reader anywhere knows what this card is waiting for. The unlock sweep greps this literal line, so without it in SOME channel nothing can ever return this card to the queue — the block outlives its blocker in silence.
H4#11810 — pm:blocked with a Blocked-by: line in NEITHER channel — not in the body, and not in any comment on the thread (both were read). Either channel discharges the duty: seats park the line in a comment on purpose, because rewriting a body through the MCP escaping hazard ([finding] GitHub MCP issue reads return HTML-escaped bodies, so any seat that appends a line by rewriting a card body may store ' literals into its code blocks #8813) is the riskier write. So this is not a formatting nit — no machine reader anywhere knows what this card is waiting for. The unlock sweep greps this literal line, so without it in SOME channel nothing can ever return this card to the queue — the block outlives its blocker in silence.
H24#11857 — pm:queue while ASSIGNED to os-elon — the board makes two contradictory claims about this one card: the queue view reads pm:queue as dispatchable NOW, and the claim protocol reads a non-empty assignee as TAKEN (⛔ never reassign). Both readers are right about what they read, so the card is available to everyone and forbidden to everyone at once — not a race, a card nobody can legally move. The measured origin is a state ROLLBACK that swapped the label and left the field: the landing re-label and the unlock scan both owe 「同笔摘 assignee」 and only dead-claim reclamation ever said so (17 carriers across three repos at the 2026-08-23 census). Remedy: whichever write set pm:queue owes the assignee drop in the SAME stroke — do it now. ⚠️ Asymmetric: an agent identity in that field is dead-claim residue and may be cleared on its evidence; a HUMAN assignment may be real ownership and ⛔ must never be cleared by an agent — take it to the maintainer. This row fires either way and states the login so the reader can tell them apart: the rule lands first and an ownership exemption is an explicit marker later, never the other way round (ruling 2026-08-23).
H30#11887 — pm:queue with ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H30#11910 — pm:queue with ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H2#11925 — assignee set but no claim comment on the thread
H34#11925 — a claim-shaped comment whose separator is NOT the canonical colon — 1 comment(s), EM DASH (U+2014) where the protocol writes Claim: — and NO comment on this thread matches the marker. So this card reads as CLAIMLESS to H2 (its row on this card is correct, not a false positive) and sits outside dispatch-liveness entirely: governingClaim gates on the same marker before reading Branch:, so H20 can never report a missing remote ref for it and H27 can never report a dead claim — the two rows that exist to catch an abandoned dispatch. Remedy, and it is the WRITE side: the claiming seat re-posts (or edits) the claim so its FIRST line begins with the literal Claim:, per SKILL.md step 4 — 「首行以字面 Claim: 开头是机器判据(维护者 2026-08-11 裁定;巡查谓词只认这一个拼写且保持严格,修法是全舰队向文档拼写收敛,⛔ 不放宽谓词)」. ⛔ The patrol predicate is NOT widened to accept the separator this card used; that is the one repair the ruling closes. Report-only: ⛔ never a label written from this script.
H22#11968 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H19#11976 — pm:blocked while 1 of 2 Blocked-by: target(s) — read from body OR comment — is CLOSED (#11970 (closed 2026-08-29T09:58:29Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. 1 target(s) are still open (#11974), so this is a PARTIAL discharge and the card may still be legitimately blocked — the row reports it, it does not decide it. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H19#11977 — pm:blocked while 1 of 2 Blocked-by: target(s) — read from body OR comment — is CLOSED (#11970 (closed 2026-08-29T09:58:29Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. 1 target(s) are still open (#11974), so this is a PARTIAL discharge and the card may still be legitimately blocked — the row reports it, it does not decide it. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H26#11978 — The wait is TRANSITIVE: #11975 is itself pm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.
H26#11979 — The wait is TRANSITIVE: #11978 is itself pm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.
H30#11995 — pm:queue with ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H19#12004 — pm:blocked while 1 of 1 Blocked-by: target(s) — read from body OR comment — is CLOSED (#11678 (closed 2026-08-25T09:08:24Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H19#12005 — pm:blocked while 1 of 1 Blocked-by: target(s) — read from body OR comment — is CLOSED (#11678 (closed 2026-08-25T09:08:24Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H2#12034 — assignee set but no claim comment on the thread
H19#12039 — pm:blocked while 1 of 1 Blocked-by: target(s) — read from body OR comment — is CLOSED (objectstack-ai/objectui#6302 (closed 2026-08-30T06:30:16Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-hold row, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned to pm:queue still carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind — ⚠️ agent identity only, a HUMAN assignment is ⛔ never cleared by an agent.
H24#12046 — pm:queue while ASSIGNED to yinlianghui — the board makes two contradictory claims about this one card: the queue view reads pm:queue as dispatchable NOW, and the claim protocol reads a non-empty assignee as TAKEN (⛔ never reassign). Both readers are right about what they read, so the card is available to everyone and forbidden to everyone at once — not a race, a card nobody can legally move. The measured origin is a state ROLLBACK that swapped the label and left the field: the landing re-label and the unlock scan both owe 「同笔摘 assignee」 and only dead-claim reclamation ever said so (17 carriers across three repos at the 2026-08-23 census). Remedy: whichever write set pm:queue owes the assignee drop in the SAME stroke — do it now. ⚠️ Asymmetric: an agent identity in that field is dead-claim residue and may be cleared on its evidence; a HUMAN assignment may be real ownership and ⛔ must never be cleared by an agent — take it to the maintainer. This row fires either way and states the login so the reader can tell them apart: the rule lands first and an ownership exemption is an explicit marker later, never the other way round (ruling 2026-08-23).
H2#12046 — assignee set but no claim comment on the thread
H2#12104 — assignee set but no claim comment on the thread
H22#12121 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
H30#12167 — pm:queue with ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it to needs-user-decision if it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.
H26#12183 — The wait is TRANSITIVE: objectstack-ai/objectui#6661 is itself pm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.
H22#12188 — card is CLOSED (closed completed) but still carries pm:dispatched — a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip the pm:* state label(s); no other write is owed, the card is already closed.
… 189 further row(s) omitted to fit GitHub's issue-body limit; the full list is in the workflow run log.
On-hold trigger-file index (H17)
Before dispatching, intersect your dispatch's file surface against this list and NAME any card it hits in the dispatch brief. These are the trigger files open holds declare — the opportunistic-restart mechanism (maintainer-accepted 2026-08-11) whose intersection was measured at 0-for-19 while it lived only as a remembered protocol step (#10034). Report-only: a card here is a hold in good standing, never a finding. Extraction is deterministic — every path shown is a tracked file; anything unverifiable was dropped rather than guessed, so this list under-reports and never invents. (read on 78 of 78 open pm:on-hold card(s); 7419 tracked file(s) in the oracle.)
os-half-state-sweep — machine-findable marker for this generated view.
Generated view — not a second tracker. Authority lives on each card and PR (one-board rule); this body is rewritten IN PLACE by the scheduled patrol workflow (
.github/workflows/half-state-patrol.yml) on every run, and the edit history is the archive. Report-only: every row is patrol input, never a gate verdict, and this sweep never fixes a state. Each predicate and the protocol clause it enforces are documented inscripts/pm/check-half-states.mjs.Swept 2026-08-30T07:42:47.894Z · run 33299794257 · commit
ececf7a216eec0128b5c97ed6462b68caed7a179· triggerscheduleThe timestamp above is the patrol's own heartbeat: a
Sweptline that stops advancing means the standing caller died, which is the failure this anchor was created to make visible. Read it before you read the rows.check-half-states: swept 281 open pm-/p0-labeled issue(s), 356 open issue(s) in the unscoped pass (H13–H15, H18), 25 open PR(s) (merge state read on 0 of 0 H16 candidate(s)) and 397 recently-merged PR(s) in objectstack-ai/objectstack — 262 half-state(s) found. H22 read 400 recently-closed issue(s) for
pm:*state residue (bounded window; older closed carriers are outside it by design). H23 read 300 squash commit message(s) from the default branch's recent window, carrying 3 closing-keyword binding(s) across 3 message(s) (bounded window; a message that landed before it is invisible by design). Hold comments read on 78 of 78 H17 candidate(s).Blocked-by:comment fallback read on 43 of 43 candidate(s).Restart-when:hold comments read on 33 of 33 H9 candidate(s). Blocker liveness (H19): targets resolved on 23 of 25 distinctBlocked-by:target(s) named by openpm:blockedcard(s) — the 2 unresolved target(s) are named on their own cards' rows, and those rows sort ABOVE the size trim so they cannot be what a truncated body drops (#11218: this clause used to be an unconditional promise, and on the 2026-08-25T02:08Z sweep it was false — 199 rows were trimmed and not one rendered row carried an unresolved target) Cross-repo reachability was measured directly on 1 sibling repo(s), of which 1 do(es) not answer this credential — those targets are unjudgeable by ruling (each install reads its own repo with its own repo-scoped token) and ⛔ no re-run resolves them. Dispatch liveness (H20 + H27): remote branch read on 7 of 7 distinct claimed branch(es) named by openpm:dispatchedcard(s) past the 60-minute threshold — one read serving both rows, so H27's 24h population is a subset of this one and costs no request of its own. Seat liveness (H32): marker thread read on 5 of 5 HELD seat post(s) whose lane is countable on THIS board — a seat held for a sibling repo's lane is out of scope here (its inventory is unreadable from this sweep, so an empty-looking queue would mean nothing), and an unread thread makes H32 decline to judge that seat rather than accuse it. Gate-removal patrol (H35): 0 removal(s) of a gate-semantic label read from 0 page(s) of the repo-wide issue-event stream over the last 12h — no per-card timeline fetch. 0 of them are UNJUDGEABLE (a gate that only ever had ONE carrier leaves 「双载体同笔清标」 no evidence in either direction, so neither H31 nor H35 can say cleared-or-stripped). Shared-file holds (H36): changed-file page read on 24 of 24 open PR(s) — a pair needs both sides read, so a shortfall can only MISS a hold, never invent one. Family folds (H37): 2 live shared branch(es) claimed by more than their own chain head, and a member comment page read on 99 of 99 openpm:queuecard(s) — that second read is bought ONLY when a fold is live, so 0 of 0 is a board with no fold in flight rather than a pass that skipped one, and an unread member can only MISS a drifted write, never invent one. Report-only: findings are patrol input, not a gate verdict.Findings
pm:blockedand 1 of 1Blocked-by:target(s) could NOT be resolved this sweep (objectstack-ai/cloud#944(HTTP 404;objectstack-ai/cloudis NOT readable to this sweep's credential)) — so whether this block has outlived its blocker is UNJUDGED, not confirmed. Unread is not still-open (check:react-declaration-parity 是唯一没接进任何 workflow 的源码审计门禁,且无 MANIFEST 时静默 skip 退出 0 —— 它现在永远不可能红 #4690): a target dropped in silence reads as a healthy block forever, which is the exact failure this item exists to end, so it is named here instead.GET /repos/<owner>/<name>— not inferred from the issue 404). That is the cross-repo class the contract-first split manufactures, and it is a standing, ACCEPTED limit rather than a defect to chase: each patrol install reads its own repo with its own repo-scoped token by ruling, so no re-run and no re-read of this card will ever resolve these. ⛔ Do not "fix" it on the card — judge the target BY HAND, or take a credential change to routing/security, whose call it is. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:blockedand 1 of 1Blocked-by:target(s) could NOT be resolved this sweep (objectstack-ai/cloud#1509(HTTP 404;objectstack-ai/cloudis NOT readable to this sweep's credential)) — so whether this block has outlived its blocker is UNJUDGED, not confirmed. Unread is not still-open (check:react-declaration-parity 是唯一没接进任何 workflow 的源码审计门禁,且无 MANIFEST 时静默 skip 退出 0 —— 它现在永远不可能红 #4690): a target dropped in silence reads as a healthy block forever, which is the exact failure this item exists to end, so it is named here instead.GET /repos/<owner>/<name>— not inferred from the issue 404). That is the cross-repo class the contract-first split manufactures, and it is a standing, ACCEPTED limit rather than a defect to chase: each patrol install reads its own repo with its own repo-scoped token by ruling, so no re-run and no re-read of this card will ever resolve these. ⛔ Do not "fix" it on the card — judge the target BY HAND, or take a credential change to routing/security, whose call it is. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:on-hold, open ~76d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card'sBlocked-by:/Restart-when:liveness in the triage round.pm:on-hold, open ~42d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card'sBlocked-by:/Restart-when:liveness in the triage round.pm:queuewith ~10d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.pm:blockingcard: open ~461h, 4 of 4 openpm:blockingcard(s) unassigned. The lane selection order putspm:blockingsecond only topriority:p0, so an unclaimed one aging while fresher cards are picked is selection-order drift — visible here by name instead of only in a seat's memory. Age is the CARD's (created_at, the same quantity the order's within-rank tie-break reads), not the label's: that would need a per-card timeline fetch this sweep never makes. Visibility row — no threshold, it reports unconditionally, and like everything here it is patrol input, not a verdict.completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.pm:on-hold, open ~17d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card'sBlocked-by:/Restart-when:liveness in the triage round.Blocked-by:line, body or comment (Triage the parameterless-catch read seams, then decide the 2a criterion — the one surviving half of #9165, on numbers that need re-measuring first #12753), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.pm:on-holdwith noRestart-when:line in EITHER channel — not in the body, and not in any comment on the thread (both were read) — the hold state is legal only with a machine-fireable exit (Restart-when: closed <owner/repo>#N, or a one-line executable predicate).-/*bullet,**bold), but a mis-spelled or lowercased key is not — the unlock scan greps the literal. If the line is there, unwrap or re-spell it; that is the whole fix, and no state change is due. Add or repair the line first. Closing is the LAST resort and applies only to a card that genuinely has no fireable exit: such a card is closednot plannedwith reason + provenance in the closing comment (type:Bug holds re-route instead — see the state model's Bug branch). Channel: likeBlocked-by:(H4/H14), aRestart-when:line counts from the body OR a comment — either channel discharges the duty, and the unlock scan reads both (hold protocol: widen theRestart-when:channel contract to body OR comment — unlock scan + H9 + the SKILL.md hold-row text move together (a comment-channel exit silently expired for 2 days) #10403 closed the old body-only gap).completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.pm:queuewith ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.needs:contract-reviewon the CARD while its delivering open PR feat(service-automation): definition-level input-schema refusal is non-retryable (FLOW_INPUT_SCHEMA_INVALID) #13371 (draft) does NOT carry it — the two carriers of one gate disagree, so the pair was written half way: either the hang never reached the PR side (「PR 一存在即挂」, and the PR exists), or a PASS cleared the PR side and stopped there, leaving the card gated behind a review that has already passed. The gate is a DUAL carrier — 「两边都挂好」, hung in one stroke and cleared in one stroke, each carrier written read-modify-write with a READ-BACK (「闸门被剥不是红灯是放行」: a stripped gate is a GREEN light, and 「被剥」 and 「从未挂过」 are indistinguishable in the evidence, so the read-back is the only way either is ever noticed). Report-only: ⛔ never a label written from this script — hanging or clearing a review gate from a sweeper would be issuing the verdict, which is 自查放行.pm:on-hold, open ~11d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card'sBlocked-by:/Restart-when:liveness in the triage round.pm:queuewith ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.pm:queuewith ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.pm:dispatched— the merge's paired write never landed. Droppm:dispatchedand re-grade the remainder (re-queue, close, or block the un-delivered half) in the same stroke, and 「同笔摘 assignee」 — the landing re-label owes the ASSIGNEE DROP too. A re-graded card that keeps the finished dev's assignee lands straight in H24's two-views contradiction (pm:queue+ assigned = dispatchable to the queue view, taken to the claim rule), which is how 17 cards across three repos got stuck where nobody could legally move them (finding(pm): open+pm:queue+assignee 双向撒谎态,三仓 17 张——四条机器门修法(新 H 规则/配对写入补 assignee/巡查跨仓/H22 窗口),外加一个需裁决的「等人工动作」状态 #11196).pm:blocked, open ~7d (threshold 7d) — the important-parked inventory exists so a bug or security card cannot age out of sight inside a parked state. Re-check the card'sBlocked-by:/Restart-when:liveness in the triage round.pm:blockedon 1 target(s) that can never CLOSE:#7497(pm:on-hold). The unlock predicate is "theBlocked-by:target closed", andpm:on-hold/needs-user-decisionare by definition states a card sits in WHILE OPEN — so this block has NO MECHANISM THAT WILL EVER RELEASE IT. Every existing check passes on this card (the line is present, the target resolves, the target is open, the label is correct), which is why the measured instances were found by a human reading and by no gauge; H9 asks the mirror question about the HELD card and nothing asked about the WAITING one.pm:on-holdwith noRestart-when:line in EITHER channel — not in the body, and not in any comment on the thread (both were read) — the hold state is legal only with a machine-fireable exit (Restart-when: closed <owner/repo>#N, or a one-line executable predicate).-/*bullet,**bold), but a mis-spelled or lowercased key is not — the unlock scan greps the literal. If the line is there, unwrap or re-spell it; that is the whole fix, and no state change is due. Add or repair the line first. Closing is the LAST resort and applies only to a card that genuinely has no fireable exit: such a card is closednot plannedwith reason + provenance in the closing comment (type:Bug holds re-route instead — see the state model's Bug branch). Channel: likeBlocked-by:(H4/H14), aRestart-when:line counts from the body OR a comment — either channel discharges the duty, and the unlock scan reads both (hold protocol: widen theRestart-when:channel contract to body OR comment — unlock scan + H9 + the SKILL.md hold-row text move together (a comment-channel exit silently expired for 2 days) #10403 closed the old body-only gap).pm:queuewith ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.Blocked-by:line, body or comment (finding:SCIMMemberReferenceSchema.typeadmits 'Group' (RFC 7643 §4.2) but stable @better-auth/scim rejects it twice over — declaration-vs-contract fork, latent (zero runtime consumers) #11820), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.Blocked-by:line, body or comment (Authenticated data requests issue ~20 DB queries in ~17 sequential legs — per-request auth/session/localization/metadata resolution has no cross-request caching, costing ~1.5s/request on remote Postgres #10757), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.priority:p0, open and unassigned, with no activity for ~124h (threshold 24h) — p0 is queue-jump priority, so a stale unclaimed one usually means no seat's declared scan scope covers the queue it sits in. Put it in the triage round brief / decision box and name its reader.Blocked-by:line, body or comment (auto-org-admin-grant resolves theorganization_adminset id by name alone (limit 1, unscoped, process-cached), so walled org-admin grants can point at the organization-less row #11670), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.pm:blockedwhile 1 of 1Blocked-by:target(s) — read from body OR comment — is CLOSED (#9689(closed 2026-08-24T14:39:13Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.pm:queuewith ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.pm:blockedwith aBlocked-by:line in NEITHER channel — not in the body, and not in any comment on the thread (both were read). Either channel discharges the duty: seats park the line in a comment on purpose, because rewriting a body through the MCP escaping hazard ([finding] GitHub MCP issue reads return HTML-escaped bodies, so any seat that appends a line by rewriting a card body may store'literals into its code blocks #8813) is the riskier write. So this is not a formatting nit — no machine reader anywhere knows what this card is waiting for. The unlock sweep greps this literal line, so without it in SOME channel nothing can ever return this card to the queue — the block outlives its blocker in silence.pm:blockedwith aBlocked-by:line in NEITHER channel — not in the body, and not in any comment on the thread (both were read). Either channel discharges the duty: seats park the line in a comment on purpose, because rewriting a body through the MCP escaping hazard ([finding] GitHub MCP issue reads return HTML-escaped bodies, so any seat that appends a line by rewriting a card body may store'literals into its code blocks #8813) is the riskier write. So this is not a formatting nit — no machine reader anywhere knows what this card is waiting for. The unlock sweep greps this literal line, so without it in SOME channel nothing can ever return this card to the queue — the block outlives its blocker in silence.pm:queuewhile ASSIGNED toos-elon— the board makes two contradictory claims about this one card: the queue view readspm:queueas dispatchable NOW, and the claim protocol reads a non-empty assignee as TAKEN (⛔ never reassign). Both readers are right about what they read, so the card is available to everyone and forbidden to everyone at once — not a race, a card nobody can legally move. The measured origin is a state ROLLBACK that swapped the label and left the field: the landing re-label and the unlock scan both owe 「同笔摘 assignee」 and only dead-claim reclamation ever said so (17 carriers across three repos at the 2026-08-23 census). Remedy: whichever write setpm:queueowes the assignee drop in the SAME stroke — do it now.pm:queuewith ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.pm:queuewith ~5d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.Claim:— and NO comment on this thread matches the marker. So this card reads as CLAIMLESS to H2 (its row on this card is correct, not a false positive) and sits outside dispatch-liveness entirely:governingClaimgates on the same marker before readingBranch:, so H20 can never report a missing remote ref for it and H27 can never report a dead claim — the two rows that exist to catch an abandoned dispatch. Remedy, and it is the WRITE side: the claiming seat re-posts (or edits) the claim so its FIRST line begins with the literalClaim:, per SKILL.md step 4 — 「首行以字面Claim:开头是机器判据(维护者 2026-08-11 裁定;巡查谓词只认这一个拼写且保持严格,修法是全舰队向文档拼写收敛,⛔ 不放宽谓词)」. ⛔ The patrol predicate is NOT widened to accept the separator this card used; that is the one repair the ruling closes. Report-only: ⛔ never a label written from this script.completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.Blocked-by:line, body or comment (platform-admin re-anchor follow-up (Choice 4B): config-anchor thesingleposture — first-user promotion becomes development-only fallback #11979, platform-admin re-anchor L6 (reap): reader census on a walled rig; organization-scope auto-org-admin-grant's resolver; stop minting org-less rows; only then reap #11978, platform-admin re-anchor L8 (cloud seam): control-plane env injection + invite-only interaction — cloud seat leg, designed blind #11977, platform-admin re-anchor L7 (docs): authorization + permission-sets pages and the operator runbook for the admin-email variable #11976, platform-admin re-anchor L5 (migration): time-boxed legacy-grant dual read, then its removal one minor later #11975), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.Blocked-by:line, body or comment (platform-admin re-anchor L6 (reap): reader census on a walled rig; organization-scope auto-org-admin-grant's resolver; stop minting org-less rows; only then reap #11978), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.pm:blockedwhile 1 of 2Blocked-by:target(s) — read from body OR comment — is CLOSED (#11970(closed 2026-08-29T09:58:29Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. 1 target(s) are still open (#11974), so this is a PARTIAL discharge and the card may still be legitimately blocked — the row reports it, it does not decide it. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:blockedwhile 1 of 2Blocked-by:target(s) — read from body OR comment — is CLOSED (#11970(closed 2026-08-29T09:58:29Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. 1 target(s) are still open (#11974), so this is a PARTIAL discharge and the card may still be legitimately blocked — the row reports it, it does not decide it. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —Blocked-by:line, body or comment (platform-admin re-anchor follow-up (Choice 4B): config-anchor thesingleposture — first-user promotion becomes development-only fallback #11979), but NOT carryingpm:blocking— a real unblocker the selection order cannot see. The label is the derived cache that makes a card outrank everything butpriority:p0; without it this card competes on age alone while the cards waiting on it cannot start. Report-only: the remedy is the triage sweep's derivation pass applying the label, never a hand-applied one.#11975is itselfpm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.#11978is itselfpm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.pm:queuewith ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.pm:blockedwhile 1 of 1Blocked-by:target(s) — read from body OR comment — is CLOSED (#11678(closed 2026-08-25T09:08:24Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:blockedwhile 1 of 1Blocked-by:target(s) — read from body OR comment — is CLOSED (#11678(closed 2026-08-25T09:08:24Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:dispatched— the merge's paired write never landed. Droppm:dispatchedand re-grade the remainder (re-queue, close, or block the un-delivered half) in the same stroke, and 「同笔摘 assignee」 — the landing re-label owes the ASSIGNEE DROP too. A re-graded card that keeps the finished dev's assignee lands straight in H24's two-views contradiction (pm:queue+ assigned = dispatchable to the queue view, taken to the claim rule), which is how 17 cards across three repos got stuck where nobody could legally move them (finding(pm): open+pm:queue+assignee 双向撒谎态,三仓 17 张——四条机器门修法(新 H 规则/配对写入补 assignee/巡查跨仓/H22 窗口),外加一个需裁决的「等人工动作」状态 #11196).pm:blockedwhile 1 of 1Blocked-by:target(s) — read from body OR comment — is CLOSED (objectstack-ai/objectui#6302(closed 2026-08-30T06:30:16Z)): the block has outlived its blocker. Nothing else here asks this question — H4 asks whether the line EXISTS, H14 asks the REVERSE index — so an expired block sits with a well-formed line, a correct label and no row anywhere: one measured card sat ~4.5h past its blocker's close and was found only by a human walking the graph, another was released only by a manual triage pass. Every target it names is closed: nothing this card declared a wait on is still running. Report-only, and the release is NOT this script's to make: the state model gives it two mechanical double-checks (pm:blocked/pm:on-holdrow, 「放行双查」) — ① release only against the condition carried by the MOST RECENT conversion comment, never an earlier blocker on the thread (a condition already spent, re-fired, reinstates an expired premise as the current one), and ② refuse to release when the card carries a MERGED PR newer than that conversion comment (the card moved on after the condition was written, so the cited fact can be true and no longer current). This row surfaces the candidate; the unlock sweep releases it — ⛔ never a label written from this script. When that release does happen, its paired write includes 「同笔摘 assignee」: a card returned topm:queuestill carrying the assignee of the seat that parked it is dispatchable to the queue view and taken to the claim rule at the same time (H24), which is the state the unlock scan was measured leaving behind —pm:queuewhile ASSIGNED toyinlianghui— the board makes two contradictory claims about this one card: the queue view readspm:queueas dispatchable NOW, and the claim protocol reads a non-empty assignee as TAKEN (⛔ never reassign). Both readers are right about what they read, so the card is available to everyone and forbidden to everyone at once — not a race, a card nobody can legally move. The measured origin is a state ROLLBACK that swapped the label and left the field: the landing re-label and the unlock scan both owe 「同笔摘 assignee」 and only dead-claim reclamation ever said so (17 carriers across three repos at the 2026-08-23 census). Remedy: whichever write setpm:queueowes the assignee drop in the SAME stroke — do it now.return res.json()methods erased toPromise< any >, and measure the 38 third-party ones #13082 (merged 2026-08-29) is MERGED but the card still carriespm:dispatched— the merge's paired write never landed. Droppm:dispatchedand re-grade the remainder (re-queue, close, or block the un-delivered half) in the same stroke, and 「同笔摘 assignee」 — the landing re-label owes the ASSIGNEE DROP too. A re-graded card that keeps the finished dev's assignee lands straight in H24's two-views contradiction (pm:queue+ assigned = dispatchable to the queue view, taken to the claim rule), which is how 17 cards across three repos got stuck where nobody could legally move them (finding(pm): open+pm:queue+assignee 双向撒谎态,三仓 17 张——四条机器门修法(新 H 规则/配对写入补 assignee/巡查跨仓/H22 窗口),外加一个需裁决的「等人工动作」状态 #11196).completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.pm:queuewith ~4d with no activity of any kind (threshold 3d) — the queue is the one state that asserts the card is dispatchable NOW with nothing left to ask, so a card sitting in it is not inventory the way a parked card is: it is a card the lane keeps passing over. The measured shape is a judgement that WAS made and never written (「判断做了(有分析产出),状态写入没做」) — the analysis lands in a comment and the state stays where it was, which is indistinguishable from an ordinary queued card at every glance. This row does not judge the card and does not rank it: it asks for ONE explicit transition — dispatch it, convert it toneeds-user-decisionif it turns out to carry an unanswered question (the queue means 「无可问之事」), park it with a machine-fireable exit, withdraw it, or rewrite a premise that no longer holds. Report-only: ⛔ never a label written from this script — choosing which of those transitions applies is the whole judgement.objectstack-ai/objectui#6661is itselfpm:blocked, so this card is waiting on a card that is waiting. A single-level predicate cannot see past one hop, and the measured chain was real one level up and FALSE two levels up (the target was an H19 finding on the same sweep — both of ITS blockers had closed). This row does not chase the chain; it says to look one level further.completed) but still carriespm:dispatched— a state label is a claim that work is in flight, and the card left the board without the paired write that clears it. H8 would have flagged this while the card was open; it closed first, which is the normal path rather than the rare one. Strip thepm:*state label(s); no other write is owed, the card is already closed.On-hold trigger-file index (H17)
Before dispatching, intersect your dispatch's file surface against this list and NAME any card it hits in the dispatch brief. These are the trigger files open holds declare — the opportunistic-restart mechanism (maintainer-accepted 2026-08-11) whose intersection was measured at 0-for-19 while it lived only as a remembered protocol step (#10034). Report-only: a card here is a hold in good standing, never a finding. Extraction is deterministic — every path shown is a tracked file; anything unverifiable was dropped rather than guessed, so this list under-reports and never invents. (read on 78 of 78 open
pm:on-holdcard(s); 7419 tracked file(s) in the oracle.)packages/drivers/driver-sql/src/sql-driver.tspackages/plugins/plugin-sharing/src/sharing-service.tspackages/plugins/plugin-security/src/security-plugin.tspackages/core/src/security/auth-gate.ts,packages/runtime/src/http-dispatcher.tspackages/services/service-analytics/src/analytics-service.ts,packages/spec/src/data/currency-fraction-digits.ts,packages/spec/src/data/field.zod.tsdocs/PLATFORM_GAPS_FROM_TEMPLATES.md,packages/spec/src/ui/view.zod.tspackages/client/src/realtime-api.ts,packages/runtime/src/http-dispatcher.tspackages/drivers/driver-sql/src/sql-driver.tsscripts/check-durability-degradation-log-level.mjsscripts/check-durability-degradation-log-level.mjscontent/docs/deployment/meta.json,content/docs/meta.jsonpackages/lint/src/data-model-rules.ts,packages/lint/src/validate-security-posture.test.tsscripts/check-cross-package-test-inputs.mjsexamples/app-todo/package.jsonscripts/check-cross-package-test-inputs.mjspackages/drivers/driver-sql/src/schema-drift.ts,packages/drivers/driver-sql/src/sql-driver.tsscripts/docs-audit/affected-docs.mjs,scripts/docs-audit/check-drift-comment.mjsscripts/pm/os-verify-lock.shscripts/durability-read-invention.baseline.jsonpackages/objectql/src/registry.tsscripts/pm/dispatch-gates.mjsscripts/check-type-check-coverage.mjsscripts/check-cross-package-test-inputs.mjs,scripts/pm/dispatch-gates.mjs