docs(changeset): retire two present-tense "in-flight PR" claims before they publish (objectui#9706)

[os-sales]

Fixes #9706

Clause-②: no

Two unconsumed .changeset/8478-*.md bodies asserted, in the present tense, that part
of their remaining scope was "held by in-flight PR ...". A changeset body publishes
verbatim into the CHANGELOG, so at the next @object-ui/types release these become
permanently published false statements about repo state. The window is live: release pull
request objectui#5400 is open and its branch already carries both sentences verbatim in
packages/types/CHANGELOG.md. Repairing the bodies on main regenerates that branch.
⛔ Nothing here touches objectui#5400 or changeset-release/main.

The two PR states, read live from the REST API before writing

Read at 2026-09-17, GET /repos/objectstack-ai/objectui/pulls/NUMBER, not taken from the
card's restatement:

PR	reading
objectui#8763	state=closed merged=true merged_at=2026-09-09T08:11:20Z
objectui#8799	state=closed merged=true merged_at=2026-09-09T11:40:18Z
objectui#8766 (control, cited in past tense already)	state=closed merged=true merged_at=2026-09-09T05:33:13Z

What changed — one sentence per file, prose only

.changeset/8478-describe-line-addresses.md, before:

The remaining 15 addresses (form.zod.ts, layout.zod.ts, complex.zod.ts) are out of
scope for this PR — held by in-flight PR #8763 (form.zod.ts /
layout.zod.ts); complex.zod.ts's hold (PR #8766) cleared when
that PR merged during this round, so its 6 addresses return to the queue rather than
riding this PR's scope.

after:

The remaining 15 addresses (form.zod.ts, layout.zod.ts, complex.zod.ts) were out
of scope for this PR — form.zod.ts / layout.zod.ts were held while PR
#8763 was open; complex.zod.ts's hold (PR
#8766) cleared when that PR merged during this round, so its 6
addresses returned to the queue rather than riding this PR's scope.

.changeset/8478-zod-pins-form-layout.md, before:

The remaining 6 addresses (zod/complex.zod.ts) stay out of scope for this PR — held by
in-flight PR #8799 — and the card does not close here.

after:

The remaining 6 addresses (zod/complex.zod.ts) stayed out of scope for this PR and
returned to the queue rather than riding this PR's scope — the card did not close here.

Neither replacement contains a present-tense assertion about any other PR, branch or card
— that is the defect class itself, and a repair that reintroduced it would rot the same
way. The objectui#8766 clause in the first file was left verbatim: it was already
past tense and accurate, and it is not this card's target.

The six complex.zod.ts addresses — the contradiction, settled

The two bodies disagreed about the same six addresses: the first said the hold cleared and
the six returned to the queue; the second (written later) said those six were held by
in-flight objectui#8799. The first is the true one, and both now say the same thing.
Evidence, in order of decisiveness:

1. The second claim was already false when it landed. objectui#8799 merged
   2026-09-09T11:40:18Z. The pull request that introduced that body, objectui#8822,
   merged 2026-09-09T12:52:14Z — 72 minutes later. The sentence was born describing a
   state that had already ended.
2. A separate PR carried the six. objectui#8835 (fix(types): drop stale source-line citations from 6 published describe() strings) merged 2026-09-09T13:31:48Z; its file
   list is exactly .changeset/8478-zod-pins-complex.md (added) and
   packages/types/src/zod/complex.zod.ts (modified). The six went back to the queue and
   were picked up on their own, which is precisely what the first body said would happen
   and the opposite of "riding objectui#8799's scope".
3. objectui#8478 reached state=closed at 2026-09-09T13:31:50Z, two seconds after
   objectui#8835 merged.
4. The tree agrees today. A paren-matching scan of every .describe() argument on
   396a1c198 finds 0 NAME.ext:NNN addresses in complex.zod.ts (167 .describe()
   calls), form.zod.ts (203) and layout.zod.ts (91). Lit control on the same corpus:
   the bare address matcher still fires 7 times in complex.zod.ts — in // and JSDoc
   comments, which were never in objectui#8478's population.

⚠️ Also checked: objectui#8799 did modify complex.zod.ts, so it was a genuine hold
while it was open. The defect is the tense, not the attribution.

Scope — exactly two files

 .changeset/8478-describe-line-addresses.md | 10 +++++-----
 .changeset/8478-zod-pins-form-layout.md    |  4 ++--

Frontmatter untouched, and that is measured rather than asserted:
check-changeset-overwrite prints declared at base: @object-ui/types: patch /
declares now: @object-ui/types: patch for both files.

Gates — exit code captured to disk before any pipe

gate	exit	printed verdict
pnpm check:changeset-claims	0	No pending changeset names a file this change touches. (report-only)
node scripts/check-changeset-overwrite.mjs	0	2 modified, both @object-ui/types: patch at base and now — the gate's own case 2, "CORRECTING a declaration on purpose ... prose that no longer matches" (report-only)
node scripts/check-changeset-presence.mjs	0	No source or published contract of a released package changed in this range, so no changeset is owed.
node scripts/check-changeset-fixed.mjs	0	All workspace packages are in the changeset fixed group.
node scripts/check-changeset-no-major.mjs	0	No changeset declares a major bump.
pnpm check:control-bytes	0	OK (scanned 7814 tracked text file(s); skipped 85 binary)
pnpm check:new-line-citations	0	0 new citation(s), enforcement report-only, with its own four synthetic controls PASSing
pnpm check:pending-changeset-literals	0	No test source names a pending changeset.
node scripts/check-governed-queue-guard.mjs --test on both paths	0	NOT GOVERNED — 2 path(s) checked against 5 governed surface(s)

Controls, because a zero on an unlit instrument is not a reading:

• Governed guard, lit: the same script on AGENTS.md exits 3, GOVERNED.
• Control-byte scan of the two files: grep -naP exits 1 (no hits); the same
  expression on a file holding one BEL byte exits 0.
• The in-flight scan below reads 0 on a corpus where the lit control reads 1327.

⚠️ The census is deliberately not a git grep: these bodies wrap at about 80
columns and the offending sentence wraps mid-phrase, so a line-anchored matcher cannot see
it. The scan collapses whitespace per file first. Before: 2 hits. After: 0, lit control
1327 files of 1637, absent control 0.

No source file moved, no released package moved, no other changeset touched. Nothing owed
in pnpm test / pnpm lint: the diff is markdown prose in .changeset/, and no package
source changed. Repository-wide scans belong to CI.

Deliberately not done

• Other pending changesets of the same shape are reported, not repaired —
  objectui#9713 already tracks that wider class and the routing is the PM's. The
  whitespace-tolerant census over all 1637 pending declarations found no other "in-flight
  PR" claim; the one adjacent observation is recorded in the report rather than acted on.
• objectui#5400 and changeset-release/main: untouched, unread beyond what the card
  already recorded, not commented on.

---

Generated by Claude Code