check-half-states: two report-only rows — pm:epic with pm:queue (reserved and handed over at once) and a PR whose Fixes target is still pm:queue or carries no Claim: matching its branch (implemented without a claim) #15667

Activity

1. os-steve

   self-assigned this

   on Sep 5, 2026
2. os-steve

   added

   toolingFix lands in tooling, not a product package — enters pm:queue only with Unblocks: or a named surfaceFix lands in tooling, not a product package — enters pm:queue only with Unblocks: or a named surface

   priority:p2Medium: important, M3Medium: important, M3

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   domain:skillsDomain lane — seat card indexed by label:pm:seatDomain lane — seat card indexed by label:pm:seat

   on Sep 5, 2026
3. claude

   added

   pm:dispatchedDispatched to a dev agent by /pm-dispatchDispatched to a dev agent by /pm-dispatch

   and removed

   pm:queueReady for the PM dispatch loopReady for the PM dispatch loop

   on Sep 5, 2026
4. claude commented on Sep 5, 2026

   

   claudeBot

   on Sep 5, 2026 – with Claude

   Contributor

   More actions

   Claim: PM loop round 5 — filed and dispatched in the same act on the maintainer's「ok」(2026-09-05); pure code, contract-tier review then ready + auto-merge
   Session: session_019RfFHiRCSs3JXLK4cwcfox
   Branch: claude/issue-15667-half-states-epic-and-claimless-pr-rows
   Worktree: objectstack-issue-15667
   Domain: domain:skills
   File surface: scripts/pm/check-half-states.mjs only — rows H45 (pm:epic + pm:queue; the parent-consistency half only at ≤1 read per pm:epic card) and H46 (an open PR closing a card that is still pm:queue or has no Claim: naming the PR's branch), both report-only, both in the state class, with self-test cases in both directions and a live run over both repos in the PR body (stop on breach; explain in the report)
   Container & model: M (two predicates in a 12k-line report-only script with its own self-test harness), mode:subagent, model: opus (dispatch-gates.mjs --tier scripts/pm/check-half-states.mjs at fa125f3bf, 2026-09-05T02:3xZ: no path-derived mandate)
   Clause-②: no
   Serial constraints cleared: no open PR touches scripts/pm/check-half-states.mjs (open-PR scan 02:28Z: zero mentions; PR #15577 landed 22:41Z yesterday); the H43/H44 precedents are on main. Not governed ⇒ the seat's contract-tier review, then ready + auto-merge (the queue's SQUASH governs) once every check is green. The PR carries Fixes #15667.

   Decision re-read (02:3xZ): the maintainer's rulings are quoted verbatim on #15666; this card is their mechanical half; no ruling owed.

   ---

   Generated by Claude Code
5. claude commented on Sep 5, 2026

   

   claudeBot

   on Sep 5, 2026 – with Claude

   Last edited by claude

   Contributor

   More actions

   os-dev-report

   {
     "issue": 15667,
     "status": "done",
     "branch": "claude/issue-15667-half-states-epic-and-claimless-pr-rows",
     "pr": "https://github.com/objectstack-ai/objectstack/pull/15690",
     "premise_still_valid": true,
     "summary": "Added H45 and H46 to scripts/pm/check-half-states.mjs, both report-only and both in the `state` band; no other file changed and exit semantics are untouched. H45 is a pure label intersection over cards the sweep already lists (`pm:epic` + `pm:queue`), H3's shape and H3's zero-request cost. H46 flags an OPEN PR (drafts included) whose body binds a closing keyword to an open card that either still carries `pm:queue` (leg a) or has no comment that is a `Claim:` whose `Branch:` line names the PR's head branch (leg b). Every marker is READ, never re-spelled: closingKeywordTargets (H7/H21), CLAIM_COMMENT_MARKER and claimedBranches (the pair check-clause2-carriers.mjs reads); no existing row's predicate or wording moved. H46 is judged AFTER H44 on purpose — the threads leg (b) buys would otherwise widen H44's corpus and change what an existing row reports. Budget held: the card side is `seen` union the unscoped open listing (already paged), the PR side is the open-PR listing, and the only request bought is one comment page per bound non-`pm:queue` open card on the sweep's existing shared cache. H45's optional parent half is LEFT OUT and its cost reported (see out_of_scope_findings). Draft PR #15690 is open against main with `skip-changeset` and `size/m`; body and label were both read back after writing. CI has reported green on the pushed head (34 runs: 26 success, 8 skipped, 0 failures). Not marked ready, not enqueued, no auto-merge, no approval — the seat reviews and flips.",
     "tests": "All exit codes captured before any pipe (`cmd > log 2>&1; EXIT=$?`); every claim below quotes the check's own verdict line. (1) `node scripts/pm/check-half-states.mjs --self-test` exit 0 — `✓ check-half-states self-test: 2223 cases pass.` (2) Gate derivation `node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack` on the rebased tree (no staleness warning; `--repo … checked against this checkout's 'origin' remote — it holds`) yielded 31 commands; all 31 ran under `bash scripts/pm/os-verify-lock.sh` (`VERDICT command-exit 0 · held the lock 771s · waited 36s`). 30 exited 0. The one non-zero is bare `node scripts/check-partof-closing-keyword.mjs` exit 2 — `NOT WIRED — neither PR_BODY nor PR_NUMBER is set`, recorded as NOT MEASURED, not as a failure; re-run with PR_BODY set to this PR's own body it exits 0 (`✓ check:partof-closing-keyword: this PR carries no Part-of/closing-keyword contradiction.`). Notable greens: `✓ dispatch-gates self-test: 1445 cases pass.`, `✓ check-closing-keyword-parity --self-test: 24 assertions, 5 mutations of the shipped parsers each driven to red.` (the guard extracts closingKeywordRe() from shipped bytes, so it certifies the keyword grammar did not move), `✓ check-nul-bytes --self-test: 75 assertions …`. (3) `pnpm lint` (`eslint . --no-inline-config`, the WHOLE repo, not a narrowing) through the same lock wrapper: exit 0, no problems printed — `VERDICT command-exit 0 · held the lock 112s · waited 347s`. (4) Control bytes: `grep -naP '[\\x00-\\x08\\x0b\\x0c\\x0e-\\x1f\\x7f]'` over the edited file — exit 1, zero hits. (5) ABLATION, three legs, each proving a new row can go red. No build step exists for this script, so there is no dist leg; the mutation was still proven ON DISK before every reading — the removed anchor's grep count fell 1→0, the injected text's rose 0→1, and `git hash-object` moved off the HEAD blob d6cad42a02647aad9037377863b4748a667195b4. Legs and results: H45's label pair mutated → 3 of 2223 red, first `H45: `pm:epic` + `pm:queue` -> finding`; H46 leg (a)'s `pm:queue` read mutated → 8 red, first `H46: a `pm:queue` card -> finding`; H46 leg (b)'s claim/branch match short-circuited → 5 red, first `H46: a card whose only `Claim:` names another branch -> finding`. Each leg was restored with `git checkout HEAD -- \"$REPO_ROOT/scripts/pm/check-half-states.mjs\"` (absolute path from `git rev-parse --show-toplevel`, HEAD named so the restore cannot read the mutation back out of a written index) under a `trap … EXIT INT TERM`; restoration is proven by `git diff HEAD` naming no file and the hash returning to the HEAD blob, NOT by the restore's exit code. (6) LIVE RUN, both boards, at commit 88d7a4f2d (the objectstack sweep started at pre-rebase db1b5a438, whose blob for this file is byte-identical — same d6cad42a0 — so both runs read the same code): objectstack H45=0, H46=17 (all leg b), coverage `31 of 31 bound card thread(s) were read`; objectui H45=0, H46=1 (leg a — draft PR #7621 binds a closing keyword to #7443, which is open and still carries `pm:queue`), coverage `6 of 6`. Both legs therefore fired on real boards, not only fixtures. Three objectstack hits were verified by hand against the API and are independently corroborated by H2 (`assignee set but no claim comment on the thread`) on the same cards #15321, #14025, #15068 — #15321's thread says `Claiming this card.` in prose, which is not the machine marker. The clean direction has a live control too: this PR's own body binds `Fixes #15667`, and #15667 carries a `Claim:` comment whose `Branch:` line names this head branch, so H46 is silent about it. SHA discipline: every reading above was taken at 88d7a4f2d. The PUSHED head is 13db40aac — a merge of current main, made because a rebase had left the local branch un-fast-forwardable and force-push is forbidden. The two are the SAME TREE: both `^{tree}` resolve to e0a69d2d5f0e34e8fc4a9c586b89fb6c5e4913ac and `git diff --stat 88d7a4f2d 13db40aac` is empty, so the readings are about the bytes that were pushed. Re-confirmed on the pushed head regardless: self-test exit 0 (`✓ check-half-states self-test: 2223 cases pass.`) and `check-nul-bytes: OK (scanned 7583 text file(s) … no raw ASCII control bytes)` exit 0. (7) CI on 13db40aac has fully REPORTED and is green: 34 check runs, 26 success, 8 skipped, 0 failures — including `Lint & Repo Gates`, all four `Type Check` jobs, `Test Core`, `Part-of PR must not also close its card`, `Governed Surface Queue Guard`, and `Live half-state sweep` (the patrol workflow itself, running this new code).",
     "mcp_calls": "0 — every GitHub read and write went through the container's REST channel (repo-scoped probe returned 200 at round open); no MCP GitHub tool was called.",
     "open_questions": [],
     "out_of_scope_findings": [
       "NOT FILED — for the seat to file, per the card's own disposition for H45's second half. Measured cost of `a pm:epic card whose parent's labels lack pm:epic` (2026-09-05, live API): the sub-issues endpoint GET /repos/{owner}/{repo}/issues/{n}/parent returns the parent issue WITH full `labels` (verified: #13890 -> parent #13658 carrying pm:epic), and a parentless card answers a documented 404 `No parent issue found` — so it is exactly ONE read per pm:epic card and the card's numeric condition IS met. It is left out on the budget condition instead: (a) the sub-issues endpoint family is called nowhere in this file, so it is a new fetch class, which ruling 4 forbids; (b) `pm:epic` is not among the seven label pages the sweep fetches, so the population does not exist yet, and folding a pm:epic page into `seen` would widen the input of H1/H2/H4/H9/H11/H24/H25/H29/H30 — a change to what existing rows report, which ruling 2 forbids. Population today: 5 open pm:epic cards in objectstack, 0 in objectui. A future card wanting this half needs a decision on whether the sweep may take a new fetch class and a separate pm:epic enumeration kept OUT of `seen`.",
       "OBSERVATION, no issue filed (it is the row's own output, not a repo defect): the 17 objectstack H46 rows and the 1 objectui row are live board state for the next patrol, listed in the PR body as PR -> card pairs. One is worth the seat's eye — PR #15626 closes card #14519 from branch `claude/issue-14478-duration-unit-in-key-name`, a branch naming a different card than the one it closes."
     ]
   }

   ---

   Generated by Claude Code
6. os-steve

   closed this as completedin #15690on Sep 5, 2026
7. claude

   removed

   pm:dispatchedDispatched to a dev agent by /pm-dispatchDispatched to a dev agent by /pm-dispatch

   on Sep 5, 2026