Skip to content

docs(north-star): the road grades a wrong-result run by Priority rule 2; rule 3 does not hold the p2 pm:blocking card that unblocks the lane's P0/P1 - #33279

Merged
os-zhuang merged 2 commits into
mainfrom
claude/project-thread-xzthx1
Nov 14, 2026
Merged

os-zhuang merged 2 commits into
mainfrom
claude/project-thread-xzthx1

Conversation

@objectstack-fleet

@objectstack-fleet objectstack-fleet Bot commented Nov 14, 2026 •

Copy link
Copy Markdown
Contributor

Requested by Leehom · project thread

Fixes #33163

domain:skills#1 · os-project-manager · session_01FEkQ5mgwViR8HMJPT3pTVT

Clause-②: no
Metadata-protocol: none

维护者速读

  • 改了什么:北极星「the road」段那句「跑通但结果错为 P2」改成指向〈Priority〉第 2 条定级,不再自带级别;〈Priority〉第 3 条加一个豁免:车道有未派 P0/P1 时,该 P0/P1 在 Blocked-by: 行点名的 p2 pm:blocking 卡照常派发;读第 3 条扣卡的两处(execution-duties.md 取卡前置、triage-duties.md 三门皆无的 tooling 卡首触即关)都加了指向豁免的说明。
  • 为什么改:您在 pm-dispatch(grading) + NORTH-STAR: grade in a fixed order — impact, reach discount, one raise, release window — with decidable tests for security boundary, core task, workaround and "a real user can hit" #32997 的两条裁决「「the road」段仍写「跑通但结果错为 P2」,与 Priority 新的「数据静默错或丢为 P1」不一致,要改」与「Priority 第 3 条「车道有未派 P0/P1 时不派 p2/p3 工具卡」,现在也会挡住替它解锁的 p2 pm:blocking 工具卡,要豁免」。
  • 风险与回滚:纯文字,4 行改 4 行;没有脚本实现第 3 条的扣卡(已 grep 核实),所以行为只经席位读规则改变;回滚即还原本 PR。
  • 先看:docs/NORTH-STAR.md 第 116 行(第 3 条豁免的措辞),再看第 11 行与 triage-duties.md 第 63 行。
  • 两分钟验证:git diff origin/main -- docs/NORTH-STAR.md,预期只见第 11 行末句改为指向 "Priority" rule 2、第 116 行多一个 except 从句。
  • 需要你做的:受管面 Tier H(docs/NORTH-STAR.md):请批准或合并。

What changes

  • docs/NORTH-STAR.md "The one measure: the road" (line 11). "one that runs through but gives a wrong result is P2" becomes "one that runs through but gives a wrong result is graded by "Priority" rule 2". The road now points at rule 2 (data silently wrong or lost is P1 subject to the reach discount; any other wrong run is P2) instead of restating a grade, so the two cannot drift again.
  • docs/NORTH-STAR.md 〈Priority〉 rule 3 (line 116). The hold gains its one exemption, stated once: "…dispatches no p2/p3 tooling card or contract-hygiene card, except a p2 pm:blocking card that such a P0/P1 names on its Blocked-by: line, which is not held because it unblocks that P0/P1". A p2 pm:blocking card that blocks only other cards is still held.
  • .claude/skills/pm-dispatch/references/execution-duties.md line 50, the take precondition that reads rule 3's hold: 「第 3 条」 becomes 「第 3 条(含豁免)」, a pointer rather than a copy. To stay inside the 120-byte line cap, 「每次取卡现读」 becomes 「取卡时现读」 (same reading: live at every take). 110 to 118 bytes, line count unchanged (190, ceiling 190).
  • .claude/skills/pm-dispatch/references/triage-duties.md line 63, the second reader of the hold: triage's first-touch close of a p2/p3 tooling card with none of the three doors while the lane's P0/P1 is undispatched. Its doors key on the blocker's own Unblocks: line, the exemption on the dependent's Blocked-by: line, so without a pointer triage would close the very card rule 3 now exempts. The line now ends 「「Priority」第 3 条豁免除外」; to pay for it inside 120 bytes, 「P0/P1 开着(= 未派发,在飞不算)时」 becomes 「P0/P1 未派时」, the same definition execution-duties.md line 51 states. 110 to 120 bytes, line count unchanged (101, ceiling 101).

Why

The maintainer's ruling on #32997 (comment 6867138933), verbatim:

  1. 「「the road」段仍写「跑通但结果错为 P2」,与 Priority 新的「数据静默错或丢为 P1」不一致,要改」
  2. 「Priority 第 3 条「车道有未派 P0/P1 时不派 p2/p3 工具卡」,现在也会挡住替它解锁的 p2 pm:blocking 工具卡,要豁免」

Out of scope, untouched: #32997's question 3 (rule 3's "no evidence cap" against the reach: admission p3). The filer's reading (a) stands until the maintainer rules.

Acceptance notes

  • Readers of rule 3's hold, swept on origin/main d4fbb4d: git grep for 「Priority」, 〈Priority〉, "rule 3", 第 3 条, 工具卡, "tooling card", "open P0" and "undispatched P0" across .claude, scripts/pm, skills and AGENTS.md. Two readers of the hold: execution-duties.md line 50 (the take precondition) and triage-duties.md line 63 (triage's first-touch close of a door-less tooling card). The first sweep missed the second, because that line names the hold by its condition and not by 「Priority」 or 第 3 条; contract review 6879988263 found it, and round 1 adds its pointer. filing-gate.md 〈定级判据〉 and triage-duties.md line 60 read the grading rules, not the hold. No scripts/pm/** code implements the hold (the tooling label's candidate-query consumer in ensure-pm-labels.sh filters on Unblocks: or a named surface, not on lane P0/P1), so no code or self-test changes.
  • ROADMAP.md line 13 says only "the first step that breaks is the current P0", which still agrees; it is not edited.
  • pm:blocking stays derived from the Blocked-by: reverse index (dispatch-runbook.md line 8); the exemption names the Blocked-by: line, so it reads the same index and needs no new label or state.
  • Serial constraint: pm-dispatch: a lane without CONTRACT_REVIEW_TIER (fable) sends its contract review to the review seat, never to triage — write the ruling into contract-review.md / lanes/review.md and refuse a contract-review knock addressed to triage #33157 (same lane, this round) may touch execution-duties.md; whichever lands second merges origin/main and re-runs the ratchet.

Verification

Head 8850480e, base origin/main d4fbb4d (round 1 adds 8850480; origin/main has not moved on these files).

Derived gates: 23 of 23, exit 0 each

node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 23 families (round 1 adds check:pm-half-states and check:pm-label-write, both read triage-duties.md). Each was run with its exit code captured before any pipe:

  • check-adr-links (scan and --self-test), check-closing-keyword-parity (scan and --self-test), check-comment-mask-corpus, plan-test-core-shards --self-test, check-harness-current --self-test, check:agent-test-spelling, check:ci-filter-parity, check:cross-package-test-inputs, check:doc-authoring, check:driver-memory-census, check:gitlink-declared, check:nul-bytes, check:pm-governed-merges, check:pm-half-states, check:pm-label-write, check:pm-skill-id-lint, check:pm-skill-ratchet, check:refd-timer-probe, check:skill-frame-sync, check:watch-hint-literal: exit 0.
  • @objectstack/lint check:doc-formula-expressions: exit 0 after building @objectstack/formula and @objectstack/lint under the verify lock (VERDICT command-exit 0); on the first head it read exit 3 (no lint dist) before that build.
  • --ran reconciliation: "23 derived famil(ies) accounted for — 23 run, 0 NOT-MEASURED (a DERIVED zero — all 23 recorded an exit code and none of them is 3)".
  • CI families are read from this PR's check-runs, not re-run here.
Ratchet, tier and governed readings
  • node scripts/pm/check-skill-line-ratchet.mjs: exit 0; execution-duties.md 190 lines (ceiling 190) and triage-duties.md 101 lines (ceiling 101), headroom 0 each; every edited line within 120 bytes. No ceiling moves. skills/** is not touched, so the skills token ratchet does not apply.
  • dispatch-gates --tier: "no path-derived mandate", the tier stays the PM's per-card call.
  • node scripts/pm/check-governed-merges.mjs --branch claude/project-thread-xzthx1: exit 3, GOVERNED, landing tier H (人合). 3 of 3 paths hit the register: .claude/** (execution-duties.md, triage-duties.md) and docs/NORTH-STAR.md. Size +4 / -4, under the 3000 threshold.

skip-changeset: docs/** and .claude/** only, published by no package.

🤖 Generated with Claude Code

https://claude.ai/code/session_01FEkQ5mgwViR8HMJPT3pTVT

… 2; rule 3 does not hold the p2 pm:blocking card that unblocks the lane's P0/P1

The road still said a step that runs through but gives a wrong result is P2,
while Priority rule 2 now grades data silently wrong or lost as P1. The road
now points at rule 2 instead of restating a grade. Rule 3's hold on p2/p3
tooling cards gains one exemption: a p2 pm:blocking card that the open P0/P1
names on its Blocked-by: line is not held. execution-duties.md's take
precondition points at the exemption.

Claude-Session: https://claude.ai/code/session_01FEkQ5mgwViR8HMJPT3pTVT
Co-Authored-By: Claude <noreply@anthropic.com>
@objectstack-fleet

Copy link
Copy Markdown
Contributor Author

domain:skills#1 · os-project-manager · session_01FEkQ5pjSPT7ui2EUcunaVb

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 1913510660385507697739bade8ad2786634d9d9
Local-runs: none
Scope: .claude/skills/pm-dispatch/references/execution-duties.md, docs/NORTH-STAR.md + none

Inputs read: card #33163 (body and both comments: the claim and the os-dev-report), PR #33279 (body, file list, net diff, +3/-3 over 2 files), and the check-runs on this head. The merge base is d4fbb4db; main has moved one commit past it (#33218) and that commit touches neither file, so the net diff against main is the PR diff as shown.

① Derived judgments

  • No accept set and no public surface moves: the diff is three prose lines in two governed files, published by no package. Right.
  • The one behaviour that changes is what a seat reads: rule 3's hold no longer covers a p2 pm:blocking card that an open P0/P1 of the lane names on its Blocked-by: line. This widens what a lane may dispatch, in exactly the direction the ruling relayed on the card asks for (item 2). Right.
  • No script implements the hold. Read on main: the tooling label's consumers in scripts/pm/ensure-pm-labels.sh key on an Unblocks: line or a named published surface, never on the lane's open P0/P1, and nothing else under scripts/pm/** reads rule 3. No code or self-test is owed. Right.
  • Governed landing: both paths are on the register (.claude/**, docs/NORTH-STAR.md), Tier H. Governed Surface Queue Guard is green on the head. Right.
  • Skill line ratchet: execution-duties.md stays at 190 lines; the edited line is 115 bytes, under the 120-byte cap. Lint · PM dispatch-gates and Lint · repo & tooling gates are green. Right.

② Semver level

skip-changeset is on the PR and is correct: docs/** and .claude/** only, nothing a package publishes. Check Changeset is green on the head.

Clause-②: no — matches the claim on the card and the PR body; the diff declares no narrowing and changes no published schema.

③ Boundary flags

④ Design judgment

docs/NORTH-STAR.md, the road (line 11). Holds. The stale "is P2" becomes a pointer, graded by "Priority" rule 2, and rule 2 on this head does grade a wrong run (data silently wrong or lost is P1 subject to the reach discount, any other run that goes wrong is P2). The quoted-section pointer follows the file's own convention (Grading reads "Features on the road"). The card's Do 1 asks that the road agree with rule 2; a pointer agrees and cannot drift again. No finding.

docs/NORTH-STAR.md, Priority rule 3 (line 116). The exemption is stated once and matches the ruling's letter. It keys on the Blocked-by: line, the same reverse index pm:blocking is derived from, so it needs no new label or state. It also reaches the real case: a blocked P0/P1 is undispatched, hence "open" under execution-duties.md line 51, hence it was holding its own unblocker. Two findings:

  • (a) should-fix — a second reader of the hold condition was missed and can undo the exemption. .claude/skills/pm-dispatch/references/triage-duties.md line 63 on main reads the same condition in the same words (「所入车道队列 P0/P1 开着(= 未派发,在飞不算)时,三门皆无的 p2/p3 tooling 卡同样关」) and closes the card at first touch. The three doors (lines 39 to 40) key on the blocker's OWN first line (Unblocks: #N), on a named published surface, or on measured fleet drag. The new exemption keys on the DEPENDENT's Blocked-by: line. A p2 tooling card that an open P0/P1 names on Blocked-by: but that carries no Unblocks: line of its own is therefore exempt from the hold under rule 3 and closed not_planned by triage under line 63: the P0/P1 still waits on its unblocker, which is the outcome the ruling removes. The card's Do 3 asks for every reader of the hold in the dispatch references to carry the exemption by pointer; the PR's sweep names triage-duties.md line 60 only and so reports execution-duties.md line 50 as the sole reader. Fix: either one pointer on triage-duties.md line 63 (a card an open P0/P1 names on Blocked-by: is not closed there), or a ruled statement that being named on that line counts as door one. This file is outside the diff and was read, not reviewed; the finding is against the completeness of the change on the hits files. If the seat prefers, it lands as a follow-up card rather than on this head, since rule 3 itself is correct as written.
  • (b) nit — the exemption is direct and same-lane only. "a p2 pm:blocking card that such a P0/P1 names" covers a card the lane's own open P0/P1 names directly. A second-level blocker (it blocks the exempt card, not the P0/P1) stays held, and so does a tooling card in one lane that unblocks a P0/P1 in another lane while its own lane has an unrelated open P0/P1. Both follow the ruling's wording (「替它解锁的」) and the PR body states the first on purpose, so this is an accepted reading, not a defect; worth one line to the maintainer if either case shows up in a queue.

.claude/skills/pm-dispatch/references/execution-duties.md (line 50). Holds. 「第 3 条(含豁免)」 is a pointer, not a copy, so the exemption stays stated once, as the Acceptance requires. Line 51 still defines 开放 as undispatched, and the take-order on lines 52 to 53 already ranks pm:blocking above p1, so an exempt card is picked in order with no further change. dispatch-runbook.md (the other file the card names) restates no part of the hold and needs no pointer.

  • (c) nit — 「本车道队列开放 P0/P1」 became 「本车道开放 P0/P1」. The north star still says "a lane's queue" and triage-duties.md line 63 still says 「车道队列」. The reading does not change (line 51 defines the term), so no effect on the contract; the wording now differs between the two references.

No must-fix. One should-fix (a), which earns a delta review of its fix or a follow-up card; two nits (b), (c).

Check-runs on this head: every run is completed; none failed, none pending. The skipped runs are path-filtered or opt-in jobs.

Implemented-by: claude/project-thread-xzthx1
Reviewed-by: session_01FEkQ5pjSPT7ui2EUcunaVb

VERDICT: PASS

Adopted by the seat 2026-11-14T15:43Z · isolated sub-agent at CONTRACT_REVIEW_TIER · Ruling-ref: Leehom card cmsg_01TenzsRopZj4jMbvjX8uz1XFhuiZmTH8VFgMmHRmpiGAb, option A, tapped 2026-11-14T15:37Z

…n; execution-duties keeps 队列

triage-duties.md line 63 is the second reader of Priority rule 3's hold: it
closes a p2/p3 tooling card with none of the three doors while the lane's
P0/P1 is undispatched, and it would close the very card the exemption lets
through. It now names the exemption as a pointer, within 120 bytes, paid by
folding its parenthesised definition into 未派. execution-duties.md line 50
keeps 队列, consistent with triage-duties.

Claude-Session: https://claude.ai/code/session_01FEkQ5mgwViR8HMJPT3pTVT
Co-Authored-By: Claude <noreply@anthropic.com>
@objectstack-fleet

Copy link
Copy Markdown
Contributor Author

domain:skills#1 · os-project-manager · session_01FEkQ5pjSPT7ui2EUcunaVb

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 8850480e0f299fbe52bb966147cf20ba2af6f85b
Local-runs: none
Scope: .claude/skills/pm-dispatch/references/execution-duties.md, .claude/skills/pm-dispatch/references/triage-duties.md, docs/NORTH-STAR.md + none

Delta review of rework round 1. Inputs read: card #33163 (body and all three comments, the round-1 os-dev-report included), PR #33279 (file list, the delta 19135106..8850480e, and the net diff against main on this head: +4/-4 over 3 files), and the check-runs on this head. The PR body still describes round 0; the tree is judged, not the body. The merge base is still d4fbb4db; main has moved past it and none of those commits touches the three files, so the net diff is the PR diff as shown.

① Derived judgments

  • Still no accept set and no public surface: four prose lines in three governed files, published by no package. Right.
  • The delta is two lines. triage-duties.md line 63 gains the exemption pointer; execution-duties.md line 50 takes 「队列」 back. docs/NORTH-STAR.md is byte-identical to the head reviewed in round 0. Right.
  • What a seat now reads differently: triage no longer closes, at first touch, a door-less p2 tooling card that an undispatched P0/P1 of the lane names on its Blocked-by: line. This is the same direction as rule 3's exemption and the ruling relayed on the card. Right.
  • No script implements the hold or the triage close (read on main in round 0, unchanged). No code or self-test is owed. Right.
  • Governed landing: all three paths are on the register, Tier H. Governed Surface Queue Guard is green on this head. Right.
  • Skill line ratchet: triage-duties.md line 63 is 120 bytes and execution-duties.md line 50 is 118, both within the 120-byte cap; neither file's line count moves. Lint · PM dispatch-gates and Lint · repo & tooling gates are green on this head. Right.

② Semver level

skip-changeset is on the PR and is correct: docs/** and .claude/** only, nothing a package publishes. Check Changeset is skipped on this head under the label.

Clause-②: no — unchanged from round 0 and still right; the delta declares no narrowing and touches no published schema.

③ Boundary flags

④ Design judgment

docs/NORTH-STAR.md. Unchanged in the delta. The round-0 judgment stands: the road's pointer to "Priority" rule 2 agrees with rule 2, and rule 3 states the exemption once in the ruling's own terms.

.claude/skills/pm-dispatch/references/triage-duties.md (line 63). Round-0 finding (a), should-fix: fixed. The line now ends 「「Priority」第 3 条豁免除外」. It is a pointer, not a copy, so the exemption is still stated once (the card's Acceptance), and the section name resolves through line 60 of the same file (北极星「Priority」). The condition lost its inline gloss, 「开着(= 未派发,在飞不算)」 became 「未派」; the meaning is the same, an in-flight P0/P1 is dispatched and so does not count, and execution-duties.md line 51 still carries the full definition. No new finding on this line.

.claude/skills/pm-dispatch/references/execution-duties.md (line 50). Round-0 nit (c): fixed. 「本车道队列开放 P0/P1」 is back, in step with the north star's "a lane's queue" and with triage-duties.md line 63. 「每次取卡现读」 became 「取卡时现读」 to pay for it; the rule is still read live at the pick, and line 57 (「全序每级现读/现算」) keeps the every-time reading. No finding.

Carried and new, neither blocking:

  • (b) nit, carried from round 0 — the exemption is direct and same-lane only. Unchanged by the delta; it follows the ruling's wording and stays an accepted reading.
  • (d) nit, new — an exempt card with no door is spared the close but still cannot be picked. Admission of a tooling card is a separate, unconditional rule: triage-duties.md line 39 (enters pm:queue only through one of the three doors) and execution-duties.md line 8 (the candidate set excludes a door-less tooling card), neither touched here. Line 8 sits in a hits file and was not named in round 0. So a blocker named on a P0/P1's Blocked-by: line that carries no Unblocks: line of its own is now kept open instead of closed, and becomes dispatchable once that line is written on it. That is consistent: the ruling lifts the P0/P1 hold, not the tooling doors, and after this delta the doors are the only thing left between such a card and dispatch. It is recorded so that the next reader does not take line 63's pointer for an admission. A one-line follow-up (being named on an open P0/P1's Blocked-by: counts as door one) is the maintainer's call, not owed by this PR.

No must-fix and no should-fix on this head. Nits alone: the round ends.

Check-runs on this head: all 48 are completed; none failed, none pending. Every run is success or skipped (the skipped ones are path-filtered, label-skipped or opt-in jobs).

Implemented-by: claude/project-thread-xzthx1
Reviewed-by: session_01FEkQ5pjSPT7ui2EUcunaVb

VERDICT: PASS

Adopted by the seat 2026-11-14T15:58Z · delta review by the same isolated sub-agent at CONTRACT_REVIEW_TIER · Ruling-ref: Leehom card cmsg_01TenzsRopZj4jMbvjX8uz1XFhuiZmTH8VFgMmHRmpiGAb, option A, tapped 2026-11-14T15:37Z

@objectstack-fleet

Copy link
Copy Markdown
Contributor Author

domain:skills#1 · os-project-manager · session_01FEkQ5pjSPT7ui2EUcunaVb

维护者速读

  • 改了什么:北极星「the road」段「跑通但结果错为 P2」改成指向〈Priority〉第 2 条定级;〈Priority〉第 3 条加豁免:车道未派 P0/P1 在 Blocked-by: 行点名的 p2 pm:blocking 卡不再被扣;两处派发参考(execution-duties.md:50、triage-duties.md:63)各加一句指针。
  • 为什么改:您在 pm-dispatch(grading) + NORTH-STAR: grade in a fixed order — impact, reach discount, one raise, release window — with decidable tests for security boundary, core task, workaround and "a real user can hit" #32997 的两条裁决(6867138933):road 段与「数据静默错或丢为 P1」不一致要改;第 3 条会挡住替 P0/P1 解锁的 p2 pm:blocking 卡,要豁免。
  • 风险与回滚:纯文字,3 文件 +4/-4;没有脚本实现第 3 条的扣卡;回滚即还原本 PR。
  • 先看:docs/NORTH-STAR.md 第 116 行(豁免措辞),再看第 11 行。
  • 两分钟验证:git diff origin/main -- docs/NORTH-STAR.md .claude/,只见第 11 行末句、第 116 行 except 从句和两条指针。
  • 需要你做的:受管面 Tier H(docs/NORTH-STAR.md),契约复审 PASS 评论 6880211131(Head-sha 8850480e0f299fbe52bb966147cf20ba2af6f85b)已在案,请批准或直接合并。
  • 席位意见:可以合。另有一点留给您以后定:被豁免的卡若是 tooling 卡且没有自带 Unblocks: 行,仍过不了 tooling 入队三门;本 PR 按裁决只解除 P0/P1 扣卡,不改三门。

席位核对

Verdict: ACCEPT, held for the maintainer · head 8850480e0f299fbe52bb966147cf20ba2af6f85b · card #33163 · reports 6879709704, 6880107712 (rework round 1) · contract review 6879988263 (PASS on round 0, one should-fix) and delta 6880211131 (PASS, nits only; CONTRACT_REVIEW_TIER; Ruling-ref for the in-seat review: Leehom card cmsg_01TenzsRopZj4jMbvjX8uz1XFhuiZmTH8VFgMmHRmpiGAb option A).

  • Shape: draft → main, Fixes #33163. docs/NORTH-STAR.md, references/execution-duties.md, references/triage-duties.md, +4/-4. GOVERNED, Tier H. skip-changeset; Clause-②: no; Metadata-protocol: none. No ceiling moved.
  • Rulings, read in the diff: road line points at Priority rule 2; rule 3 carries the exemption once; both readers of the hold point at it (the second reader, triage-duties.md:63, was found by the contract review and fixed in round 1).
  • Gates: 23 derived, 23 exit 0, --ran derived zero. Check-runs on the head: 48, all success or skipped.
  • Accepted deviations: the exemption is direct and same-lane only (ruling's wording); no HTML signed trailer from the dev (seat signed the body on its patch).
  • Out of scope: whether being named on an open P0/P1's Blocked-by: line should count as tooling door one — a maintainer call, not owed here; pm-dispatch(grading) + NORTH-STAR: grade in a fixed order — impact, reach discount, one raise, release window — with decidable tests for security boundary, core task, workaround and "a real user can hit" #32997 question 3 untouched.
  • Landing: Tier H. The seat does not flip this PR ready or enqueue it. It needs the maintainer's merge, or an authorised APPROVED review after which the seat lands it.

@objectstack-fleet objectstack-fleet Bot added the pm:awaiting-merge PR only: landing waits on the maintainer's own review and merge (SKILL.md 复核 ②); never on an issue label Nov 14, 2026
@os-zhuang
os-zhuang marked this pull request as ready for review November 14, 2026 16:19
@os-zhuang
os-zhuang added this pull request to the merge queue Nov 14, 2026
Merged via the queue into main with commit 348637a Nov 14, 2026
54 checks passed
@os-zhuang
os-zhuang deleted the claude/project-thread-xzthx1 branch November 14, 2026 17:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

documentation Improvements or additions to documentation pm:awaiting-merge PR only: landing waits on the maintainer's own review and merge (SKILL.md 复核 ②); never on an issue size/xs skip-changeset PR has no user-facing published change; bypasses the changeset gate

Projects

None yet

3 participants