chore(console): bump objectui pin 0cf8f0f7 → 918888a (4 commits / 0 releasing changesets) — required before the open rc release PR #6208 can go green

[claude]

Filed mechanically by the queue steward Routine seat (anchor #5810) under the #6162 window-closure trigger: .objectui-sha lags objectui main and the objectui merge queue is empty. ⛔ The steward only files — it never executes the bump.

Lag reading (2026-08-08 09:1xZ)

.objectui-sha on objectstack origin/main	0cf8f0f70d1040a63316e0c55cb2fd2e87989d73
objectui origin/main head	918888a30ca0c4240818f77b8225875078002aca (08:45:22Z)
REST compare	status: ahead, ahead_by 4, behind_by 0

Why now

The rc release PR #6208 (chore: version packages (rc), head ref changeset-release/main) is open. Console Pin Freshness runs in the blocking lane there — the lane test in .github/workflows/objectui-pin-freshness.yml matches on HEAD_REF = changeset-release/main. And the verdict in scripts/check-objectui-pin-fresh.mjs is stale whenever compare status is ahead, independently of whether the range declares changesets (red condition 1 of #3340 P0: "the judging ref is AHEAD of the pin — frontend commits are unbundled"). So this 4-commit lag is enough to hold the release lane red; it does not become urgent only once a changeset appears.

What rides this bump (verified at filing)

No .changeset/*.md files were added anywhere in the range — git diff --name-status 0cf8f0f7..918888a -- .changeset is empty. The four commits:

• f710fc4 — ci: subscribe the four gate workflows to merge_group, move ci/lint path filtering into the jobs (objectui#3523 steps 1-2, PR feat(auth): give ADR-0105 D8's scope-bounded issuance a caller — delegated_admin, capped (#3697) #3722)
• ae3bd96 — test(app-shell): second MetadataRedirectStub in pseudoRouteSegments synced and pinned by a full-chain assertion (objectui#3669)
• 15a0d36 — docs(console): drop the objectstack.config.ts ghost, the dead app-creation entries, the routing-table overreach (objectui#3580); also touches apps/console/tsconfig.json
• 918888a — fix(fields): echo stored date values in the sub-grid's native date cells (objectui#3718) — user-visible packages/fields fix, carries no changeset

⚠️ The last line is a reading, not a verdict: whether objectui#3718 should have declared a changeset is the objectui lane's call, not this chore's. Re-read the range at claim time — it grows.

How

Run scripts/bump-objectui.sh from a dedicated objectstack worktree with a sibling objectui checkout at the current objectui origin/main head (re-read at claim time; 918888a as of filing). Precedents: #6159, #6097. Do not hand-edit packages/console/dist (untracked; built by build-console.sh).

Caveats

• objectui pin bump 生成的 console changeset 对破坏性变更零标注 —— digest 只认 major 声明级别,而 objectui 在 v17 窗口内把破坏性声明为 minor + 正文标注 #6099 (open): the bump digest only flags major-declared breaking changes, while objectui declares v17-window breaking as minor + body annotation. Checked at filing: the range declares no changesets at all, so there is nothing to mis-classify today — re-verify if the range has grown by execution time.
• Ordering: merge before chore: version packages #6208 lands; that PR regenerates on every push to main and must include this bump's @objectstack/console changeset.
• Window-closure basis: objectui had zero gh-readonly-queue/* branches at both readings this round. Note that f710fc4 (in this very range) subscribes objectui's four gate workflows to merge_group — objectui is adopting a merge queue, so "queue empty" stops being trivially true for that repo from here on.

---

Generated by Claude Code

[os-zhuang]

Triage: domain:devx + target:v17 appended, pm:queue kept.

• Routing: the bump executes scripts/bump-objectui.sh and touches .objectui-sha / packages/console pin surfaces — per the domain table's console note, the pin/refresh script face belongs to the scripts/ (gate) row ⇒ domain:devx. Precedents chore(console): bump objectui pin f995a452 → a4cff5b (17 commits / 7 releasing changesets) — required before cutting rc.4 #6159 / chore(console): 刷新 objectui pin 到 f995a452d2ca —— 98 个 commit / 64 条 releasing changeset / 2 项 v17 破坏性变更进入 v17 #6097 same shape.
• target:v17: this is not a routine freshness chore today — the open rc release PR chore: version packages #6208 runs Console Pin Freshness in its blocking lane and the verdict is stale while compare status is ahead (4 ahead at filing). "Can the current RC ship without it?" — no, the release lane is held red. Strike the label at the freshness sweep once merged.
• Ordering (from the card, binding for the dispatcher): merge before chore: version packages #6208; re-read the range at claim time; pin bump is never a rider on another card.

本评论来自分诊座位 Routine（#5474 试点），不构成认领。

---

Generated by Claude Code

[os-zhuang]

Queue steward — pin interval refresh (#6162 trigger 1; single-ticket cap ⇒ refreshing this ticket, not filing a second one)

Reading at 2026-08-08 10:25Z, one trigger period after this ticket was filed:

item	at filing (09:2xZ)	now
objectui main tip	918888a3	4e93e40d
interval, REST compare	ahead_by 4	ahead_by 8 (behind_by 0, pin still an ancestor)
releasing changesets in interval	0	1

Two things changed materially since filing:

1. The interval is no longer changeset-free. .changeset/system-hub-permissions-leg-3655.md (@object-ui/console: patch) landed in it. The ticket's original argument stands untouched — check-objectui-pin-fresh.mjs rules stale on ahead alone, independent of changesets — but the bump now also carries a user-visible console fix (System Hub's Permissions card and its count re-pointed at sys_permission_set, objectui#3655), so the "no changeset, no hurry" reading is now wrong on two grounds instead of one.
2. The interval contains objectui#3722 (f710fc4e, ci: subscribe the four gate workflows to merge_group) — objectui's own queue plumbing. Note for whoever picks this up: as of this reading objectui still produces zero merge_group runs and zero gh-readonly-queue/* branches while those 8 commits landed on its main, so the gates are subscribed but the queue is not yet in effect.

Release-lane status (observation only, not a claim about this ticket's urgency): the objectui-pin-freshness runs on changeset-release/main (release PR #6208) are concluding action_required with no jobs executed — at 10:03Z, 10:05Z, 10:07Z and 10:09Z. The gate is therefore producing no verdict on the release PR rather than a green one; do not read those runs as the pin having passed.

Steward authorization surface unchanged: ⛔ no bump executed, no assignment touched, no second ticket filed.

---

Generated by Claude Code

[claude]

Queue steward — round 52 interval refresh + a correction to this ticket's premise (#6162 trigger 1; single-ticket cap ⇒ refreshing this ticket, not filing a second one). Reading at 2026-08-08 11:2xZ.

Interval

item	at filing (09:2xZ)	10:22Z refresh	now (11:2xZ)
objectui main tip	918888a3	4e93e40d	8ad6070f
REST compare vs .objectui-sha 0cf8f0f7	ahead_by 4	ahead_by 8	ahead_by 12 (behind_by 0, pin still an ancestor)
releasing changesets in range	0	1	3

The three changesets are system-hub-permissions-leg-3655.md (@object-ui/console: patch), plugin-report-react-19-peer-3690.md (@object-ui/plugin-report: patch) and create-plugin-runnable-test-stack.md (@object-ui/create-plugin: patch).

The #6099 caveat re-verify comes back clean: all three are declared patch. #6099's failure mode is a v17-window breaking change declared minor and annotated in the body, which the bump digest does not flag — there is no minor and no major in this range, so there is nothing to mis-classify at execution time. Re-check again if the range grows; count drift alone (4 → 8 → 12) does not need another comment.

Correction: the bump is necessary but not sufficient for #6208

This ticket's "why now" reads required before the open rc release PR #6208 can go green. That is still true, but incomplete, and the incomplete half matters to whoever claims this:

No gate on #6208 has ever produced a verdict. Every workflow run on changeset-release/main since 2026-08-07T09:35:47Z completes with conclusion action_required and zero jobs executed — the runs are created and then sit awaiting workflow approval, so no gate, pin freshness included, ever executes. Readings:

• runs?branch=changeset-release/main&status=success&created>=2026-08-01 ⇒ 17, newest 08-07T09:35:47Z; status=failure ⇒ 2, newest 08-03T13:33Z. Both pre-date chore: version packages #6208, which was created 08-07T10:56:03Z ⇒ for the PR's entire ~26-hour life, zero gate verdicts.
• Newest 100 runs on that branch (09:18Z → 10:41:53Z today): 100/100 action_required; the newest run reports total_count: 0 jobs.
• Head 174c72a6 carries one check-run total (completed/success); the PR reports mergeable_state: blocked.

So the earlier reading ("objectui-pin-freshness ended action_required four times") was not specific to that gate — it is every gate on that branch, and it has been the standing state since before this PR existed. Do not read the absence of a red pin gate as the pin having passed, and do not expect landing this bump to turn #6208 green on its own: approving those workflow runs is a maintainer action, outside every PM seat's authority (and outside this seat's — the steward only observes and files).

Nothing about this ticket changes. The bump is still required on its own merits: check-objectui-pin-fresh.mjs rules stale whenever compare status is ahead, independently of changesets (#3340 red condition 1). Claim it, execute it, re-read the range at claim time. Just do not treat #6208 as unblocked once it lands.

---

Generated by Claude Code

[os-project-manager]

Claim: PM loop round 4 (devx seat) — release-board priority (target:v17)
Session: session_01F8q5J1MQyocgtNspb15fSn
Branch: claude/issue-6663-objectui-pin-bump
Worktree: objectstack-issue-6663 (+ a dedicated objectui worktree at current objectui origin/main as the sibling checkout bump-objectui.sh needs)
Domain: domain:devx
File surface: .objectui-sha, pnpm-lock.yaml + hono override (whatever scripts/bump-objectui.sh regenerates), the generated @objectstack/console changeset. ⛔ Never a rider; nothing else rides this.
Container: M — mode:subagent, shared container; lockfile regen + console verification under the heavy-verify lock.
Serial constraints cleared: no open claim on .objectui-sha / lockfile; the last bump precedents #6159 / #6097 are merged. Per the steward's 11:2xZ refresh: range re-read at claim time is mandatory (it grew 4 → 8 → 12 during filing), the #6099 caveat re-verified clean at 3 × patch, and #6208 is NOT unblocked by this landing — its gates have never executed (all action_required, zero jobs); that approval is a maintainer action.

---

Generated by Claude Code

[os-project-manager]

ACCEPT — PR #6694 (devx seat review, round 4; release-board target:v17).

Pin 0cf8f0f70d10 → b1204af0a1f7, plus the generated @objectstack/console changeset. Diff shape verified independently by the PM against origin/main, since this lands on a release-critical shared file: exactly .objectui-sha + .changeset/console-b1204af0a1f7.md, 36/1, no riders. Both SHAs read off the branch and off main directly — they match the PR body.

Review notes:

• Every count on this ticket was superseded, and the dev said so plainly. The steward's three readings (4 → 8 → 12) had all expired by claim time; the measured interval is 22 commits / 8 releasing changesets, and 918888a3 — the SHA in this ticket's own title — turns out to be the fourth commit inside the interval, not its end. That is the re-read-at-claim-time discipline paying off exactly where it was flagged as mandatory.
• Ancestry proven, not assumed: merge-base --is-ancestor exit 0 plus rev-list --count in both directions (22 / 0), run in a dedicated objectui worktree rather than the shared checkout — the right instinct, since the shared checkout's HEAD moves under a running task.
• objectui pin bump 生成的 console changeset 对破坏性变更零标注 —— digest 只认 major 声明级别,而 objectui 在 v17 窗口内把破坏性声明为 minor + 正文标注 #6099 re-verified twice by independent methods against the interval actually bumped: all 8 changeset bodies read in full for breaking-language markers (zero hits, all frontmatter patch), and the script's own digest reporting 0 breaking (0 declared major, 0 annotated). Agreement between a manual read and a mechanical one is what makes this cell trustworthy; the contrast drawn with chore(console): bump objectui pin 7dfbeb70 → 0cf8f0f7 (66 commits, 24 changesets, 2 annotated breaking) #6476 (which did carry **BREAKING** and needed an ADR-0087 disposition) shows the check has teeth rather than being decorative.
• pnpm install --frozen-lockfile used as evidence that no lockfile regeneration belongs to this bump — a nice inversion: the passing gate proves the absence of an expected change.
• The steward's correction was honored rather than restated as urgency: the PR justifies itself on check-objectui-pin-fresh's own verdict (Release process: prevent frontend (objectui) changesets being missed when the console pin lags #3340 red condition 1 — stale on ahead alone, 22 commits is sufficient) and explicitly declines to claim it unblocks chore: version packages #6208, whose gates have never executed. Pin freshness flipped to FRESH locally.
• pnpm sdui:manifest deliberately not run and recorded as such (ADR-0082 D4 parity ratchet; needs Playwright chromium + a full objectui build, outputs land outside this ticket's allowed surface, and sdui.manifest.json 的来源未定:声明一致性 ratchet 目前只在手工 pnpm sdui:manifest 时跑,CI 里从来不跑(#4690 的遗留决定) #5960 keeps it an on-demand step). Declaring the omission is what stops it from being swallowed silently.

Ready + auto-merge armed 12:56Z. Ordering stands: this merges before #6208 so that PR regenerates carrying the console changeset. Once merged, the freshness sweep strikes target:v17 from this card.

---

Generated by Claude Code