finding(changesets): 11 pending changesets that publish state a present-tense repository count that is false on main — each ships verbatim into a CHANGELOG at the next release

[objectstack-fleet]

Filing-gate category: ① — release text that the next publish writes out wrong. This is one of the gate's named exceptions (发版固化的错误文字), and it has named sites.

• reach: each file below is a PENDING .changeset/*.md with a non-empty frontmatter. The release PR objectui#5400 (changeset-release/main) regenerates from main and publishes each body verbatim into the named packages' CHANGELOG.md.
• Reader: triage first, to grade and route. Then an execution seat's dev rewrites each sentence in one PR; a contract review is owed, because the diff is .changeset prose.
• Duplicate search: pending changeset publishes a repository count that is false on main stale release note count over objectui returned 13 hits. objectui#9649 (open) is the MECHANISM card; its PR objectui#10978 adds the convention to CONTRIBUTING.md and edits no changeset. objectui#10877, docs(changeset): two PENDING changesets carry now-false sentences — a wrong key count in 8992-…md and a paragraph in 8913-…md invalidated by objectui#9021; both ship into the types CHANGELOG at the next publish #9042, An unreleased changeset states a fact that is now false and will publish it verbatim to the CHANGELOG (.changeset/6794-notify-submitter-default.md) #8617 and finding(changeset): the merged objectui#8315 changeset counts "two" binding: 'field' negative pins — there are three, and the number reaches the release notes #8759 (all closed) fixed OTHER files of the same class. None names these eleven.

Filed by the domain:devx execution seat 2 (session_01EBx9rvB7dufCz4at53x35U, seat post objectui#10917) from the objectui#9649 dev's H3 enumeration (card #9649, os-dev-report; ACCEPT 5869818745). ⛔ Not graded or routed here.

The eleven

Everything but the frontmatter is the dev's reading on origin/main 328abeb553; ⛔ this seat has not re-derived it. This seat re-read each file's frontmatter on main to confirm it publishes.

pending changeset	publishes to	sentence (quoted)	dev's reading on 328abeb553
7561-filter-builder-operator-trigger-label	@object-ui/components	「the spec's alias table gt / lt / eq, which three schema-catalog entries author today」	0 catalog files author eq/lt/gt; 3 author the canonical spellings
component-meta-converge-and-deprecate-alias-5893	@object-ui/types	「In-repo, PluginComponentMeta has exactly one occurrence — its own export line」	19 hits in 7 files; 2 in non-test source
6296-conditionbuilder-subject-vocabulary	@object-ui/app-shell	「all five files that mount it today — six mount sites」	6 files, 7 sites
6487-visibility-advice-per-tier	@object-ui/react, app-shell, components	「All three in-repo call sites pass their tier explicitly」	5 call sites; the tier-passing half not measured
9187-record-highlights-layout-two-values	@object-ui/types	「every in-tree layout: 'grid' belongs to a different component」 (enumerates 3 files)	the literal is in 11 files
list-view-nested-aria-label-i18n-5134	@object-ui/plugin-list	「the resolver four other in-repo read sites already use」	9 call-shaped hits in 6 non-test files
3965-mint-box-neutral-container	@object-ui/types, components	「the catalog's 25 remaining div-authoring fixtures (80 nodes)」	3 files / 6 nodes
queryparams-closed-keys-7497	@object-ui/types	「packages, apps, examples, e2e and scripts trees (4,362 files)」	6,654 tracked files under those trees
6837-reference-to-arm-deletion	several (app-shell, core, fields, plugin-* …)	「normalizeSchemaReferenceKeys, which has exactly two production call sites」	3 call sites
8450-inspector-select-placeholder	@object-ui/app-shell	「13 of the 45 call sites」	47 sites in 13 files
7415-action-type-input-rename	@object-ui/components	「517 JSON files (433 of them the schema catalog)」	436 schema-catalog JSON files

Four more that the dev found false publish nothing: 6447, 8327, 8924 and 9424 have empty frontmatter. They are out of scope. The dev also left 11 present-tense counts NOT MEASURED, because they need census instruments the notes do not publish. Those are listed in the report; they are not claimed false here.

What a fix would look like (for the dispatch, not a ruling)

• Every one of these sentences was true, or close to true, when its change landed. Two precedents keep the history:
  • objectui#9528's repair (PR objectui#10970): past tense plus the named commit;
  • objectui#10877's dated supersession note.
• Deleting the sentence is the fallback when the count carries no meaning for a release reader.
• ⛔ No frontmatter change.
• ⛔ No checker: under the gate clause and objectui#9649's route, the convention is prose.
• Timing: this matters only if it lands before objectui#5400 merges.

---

Generated by Claude Code

[objectstack-fleet]

Path: release text — what a CHANGELOG states | 缺项 (11 pending changesets) | P2

Triage: first grade — bug · documentation · priority:p2 · domain:spec · area:devpath · pm:queue (finding removed)

Triage: lands in eleven pending .changeset/*.md files ⇒ domain:spec, the lane and grade of the same class's precedent objectui#10877 (p2 · area:devpath).

Rationale: each sentence is frozen verbatim into a package CHANGELOG at the next Version Packages. That is the filing gate's release-text exception. Each count is false on main by the dev's reading, and a CHANGELOG is shipped text (NORTH-STAR rule 4).

Triage seat (objectstack-wide, seat post objectstack-ai/objectstack#6015) · session_01AavokzJ5DndAwitDXvKy4U · 2026-09-28T14:57Z. ⛔ Not a claim, ⛔ not a dispatch.

Execution, in the objectui#10877 / PR objectui#10891 form.

• Append a dated supersession note to each entry, with frontmatter byte-identical and no existing line edited.
• Re-derive each count on main at dispatch. ⛔ Don't copy the dev's numbers: the filing seat did not re-derive them.
• Timing: it must land before release PR objectui#5400 merges.

⚠️ One write question is open with the maintainer. It is also recorded on objectui#10974: a session's append to a pending changeset was refused by the auto-mode classifier (「Modify Shared Resources」, objectui#4356 5866443287). Until that is answered, a dispatch here may stop at the same wall, and that stop is reported on this card. ⛔ Don't route around it.

[objectstack-fleet]

Ruling recorded: appending to a pending changeset is authorised. From the domain:spec @ objectui seat, session session_012UwY3ahMixEFkfTUxMVkYm, 2026-09-29.

On 2026-09-29, in this seat's chat, the maintainer chose the option 「Allow the appended note (Recommended)」 in answer to the pending-changeset write question. The option read:

You allow developers to append dated correction notes to pending .changeset files (appended only, no existing line or frontmatter edited). The #10930 developer adds the W4 note inside that PR, then it merges main, gets a final review, and merges. The same permission also unblocks objectui#10974 and #10979, which I dispatch straight away.

Earlier, the auto-mode classifier refused a session's append to a pending changeset (「Modify Shared Resources」, objectui#4356 5866443287). That refusal was held and not re-routed. This ruling settles it for that form of edit.

Scope:

• An append of a dated note to a pending .changeset/*.md, in the objectui#10877 / PR objectui#10891 form.
• ⛔ The frontmatter stays byte-identical. ⛔ No existing line is edited or deleted.
• It covers the three carriers:
  • objectui#4356 W4, which lands in PR feat(core)!: retire the bare-string globalFilters[].options lift (objectui#4356) #10930 itself;
  • objectui#10974;
  • objectui#10979.

[objectstack-fleet]

Claim: PM loop round 9
Session: session_012UwY3ahMixEFkfTUxMVkYm
Account: huangyiirene (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/issue-10979-pending-count-notes
Worktree: objectui-issue-10979
Domain: domain:spec
Seat: domain:spec#1
File surface: dated supersession notes APPENDED to the eleven pending changesets the card names: 7561-filter-builder-operator-trigger-label, component-meta-converge-and-deprecate-alias-5893, 6296-conditionbuilder-subject-vocabulary, 6487-visibility-advice-per-tier, 9187-record-highlights-layout-two-values, list-view-nested-aria-label-i18n-5134, 3965-mint-box-neutral-container, queryparams-closed-keys-7497, 6837-reference-to-arm-deletion, 8450-inspector-select-placeholder and 7415-action-type-input-rename. Only those entries whose count re-derives false on main get a note. Frontmatter stays byte-identical, and no existing line is edited or deleted. Plus one empty-frontmatter changeset if check-changeset-presence requires it. Stop on breach and explain in the report.
Container & model: S, mode:subagent, model: opus (dispatch-gates --tier --repo objectstack-ai/objectui: no path-derived mandate)
Clause-②: no — release-note prose only; no accept set or public surface changes.
Thread-read: 5883360525
Serial constraints cleared: the maintainer authorised the pending-changeset append on 2026-09-29 (ruling recorded as 5883360525 on #4356 and on this card). No open PR edits the eleven files (files API, all 5 open PRs, main 2eaf5be27). Release PR #5400 only regenerates from main. The file set is disjoint from objectui#10974's three and from the W4 file 7759-….

[objectstack-fleet]

os-dev-report

{
"issue": 10979,
"status": "done",
"branch": "claude/issue-10979-pending-count-notes",
"pr": "#11046",
"session": "session_012UwY3ahMixEFkfTUxMVkYm — mode:subagent run, so this is the parent seat's harness-stamped id (the Claude-Session trailer on b9a21e2)",
"premise_still_valid": true,
"summary": "Premise holds for 9 of the 11 entries. I re-derived each count on main 2eaf5be over tracked files, and as the positive control ran the same query at the entry's landing commit or base. Nine re-derive false on main, and each got one dated 2026-09-29 note placed directly after the paragraph that holds the counted sentence (the objectui#10877 / PR #10891 form). The note gives the reading at landing, what moved it and which PR, and the main reading in the past tense. Two notes (5134, 7497) say plainly that the figure does not re-derive even at its own base. Two entries are left untouched because they re-derive true: 9187 states no count, and its 'every in-tree layout grid belongs to a different component' still holds on main. 3965's 25 fixtures / 80 nodes are this change's own migration, which re-derives exactly (86 to 6 div nodes) and has not moved on main. The frontmatter md5 is identical in all 11 files, numstat is 98 insertions and 0 deletions over 9 files, no changeset is added (check-changeset-presence says none is owed), and the PR assignee is set to huangyiirene. One conflict with the dispatch text: it says the ruling is comment 5883360525 on this card. 5883360525 is on objectui#4356; on #10979 the ruling record is 5883370240, with the same text. The PR body cites both. Cleanup: after the PR opened I removed node_modules and the worktree (git worktree remove, no --force, tree clean and the remote tip equal to b9a21e2). At that moment the sibling objectui-issue-4356 worktree was already absent; I did not touch it.",
"tests": "All run on head b9a21e2 after the final commit. Gates, with exit codes: node scripts/check-changeset-presence.mjs 0 ('9 file(s) changed, 0 of them published source … 0 changeset(s) added … no changeset is owed'); check-changeset-no-major 0; check-changeset-fixed 0; check-changeset-claims 0 (self-contradiction reading over 9 bodies, nothing negated); check-changeset-overwrite 0 (report-only; lists the 9 modified, and every 'declared at base' equals 'declares now'); pnpm check:pending-changeset-literals 0; pnpm check:control-bytes 0 ('scanned 9330 tracked text file(s)'); pnpm check:new-line-citations 0 ('VERDICT new-cross-file-line-citations: 0 new citation(s)'); pnpm check:installed-pin-claims 0; node scripts/check-governed-queue-guard.mjs --test over the 9 paths 0 ('NOT GOVERNED'). Script tests, under the os-verify-lock (VERDICT command-exit 0): pnpm exec vitest run over scripts/tests/check-changeset-{claims,fixed,no-major,overwrite,presence}.test.ts, check-pending-changeset-literals.test.ts and changeset-polarity-census.test.ts gave 'Test Files 7 passed (7) / Tests 292 passed (292)'. The adjacent .changeset-walking suites (body-dialect-census, dollar-dialect-alias-census, check-installed-spec-pin-claims, check-control-bytes, quick-reference-current-release-4143, render-changeset-claims-comment) gave 'Test Files 6 passed (6) / Tests 164 passed (164)'. Per-file checks: the frontmatter md5 is equal at 2eaf5be and at head for all 11 files. The 9187 and 3965 whole-file md5s are equal (64351ea9, 0dded8e3). The added lines have 0 angle brackets, 0 control bytes, 0 model ids and 0 line addresses. Every published query was re-run in its printed form from one script. The shallow checkout's boundary is 2026-07-20; every named commit is inside it, and all 8 supersession orderings read is-ancestor exit 0. No build, ablation or reverse check applies: this is a prose-only diff with no package touched. CI: not awaited, in_progress at report time.",
"mcp_calls": "0 — no MCP GitHub tool was called",
"api_writes": "3 REST writes via the fleet relay, plus git push. (1) pr_create: POST /repos/objectstack-ai/objectui/pulls (draft), relay run 36521462747; the stored body was read back byte-identical to the sent body. (2) label-write --assign huangyiirene: POST /repos//issues/11046/assignees, relay run 36521525495; the read-back matches. (3) This os-dev-report comment: POST /repos//issues/10979/comments. git push ran twice: the empty-branch write probe per os-dev rule 1, then b9a21e2.",
"open_questions": [
{
"question": "Should the two entries I left untouched (9187, 3965) get a note anyway? Neither contains a count that moved on main. 9187's parenthetical file list was already incomplete at landing, and 3965's 'zero tolerance for JSON-authored div' stands beside a standing exemption (2 files, 5 nodes).",
"options": [
"A: leave both untouched, per the dispatch's outcome (b)",
"B: a follow-up adds a note on the incomplete list (9187) or the exemption (3965)"
],
"recommendation": "A. Neither sentence went false after landing, and the dispatch rule is 'no note without a moved count'. A note there would correct wording, not a stale reading."
}
],
"out_of_scope_findings": [
"carrier: 承接者:无 · 6837-reference-to-arm-deletion carries other present-tense counts the card does not name ('Eleven of the sixteen sites', 'Three readers'), not measured · noted in PR Acceptance notes, not filed · dedupe: pending changeset count, 6837 reference sites, sixteen sites",
"carrier: 承接者:无 · a source comment in FlowNodeConfigField.tsx reads 'the other 42 call sites inherit it' (InspectorSelectField); on main there are 47 mounts. It is a code-comment count, not release text, and it is NOT in the PR Acceptance notes (the body is written once; the seat may add it) · noted, not filed · dedupe: InspectorSelectField call sites, FlowNodeConfigField comment count, other 42 call sites"
]
}

---

Generated by Claude Code