docs: cite the landing commits where ten objectui issue links answer 404 (objectui#10755)

[objectstack-fleet]

Fixes #10755
Clause-②: no

Dispatched implementation of the domain:ui seat 2 claim (comment 5853230223) on objectui#10755, session https://claude.ai/code/session_014mXUNuFomfj24w7s1pZzhN. Citations only, on three published docs pages: no sentence's claim moves, and the frontmatter of all three files is byte-identical.

What changed

• The ten objectui issue links that answer 404 now cite the landing commit on main, as a 9-character backticked sha. That is the form content/docs/guide/ci-cd-pipeline.md already uses for commits (five of its six existing sha citations are 9 characters), and the form PR objectui#10707 chose for the same class on objectui#10701.
• Two prose-only mentions of the same dead numbers, inside the same two sentences, are re-pointed so each sentence stays coherent: "in finding(react/plugins): useNavigationOverlay is handed a bare schema.objectName by three components that resolve their record source through the #6939 ladder — so the record-page URL can name a different object than the rows #7638's card body" reads "in the second card's body", and "over the file metadata-admin's object designer still PUTs a half-filled lookup — the surface the objectui#7714 dogfood actually wedged is neither writer the ruling named #8057 is entirely about" reads "over the file that third writer lives in". Lychee does not read those; the reader does. Each sentence's claim is unchanged.
• Commits, not PRs, at every site. Two of the landing PRs answer 404 themselves (PR 7637 for objectui#7627, PR 7805 for objectui#7753), so PR numbers are not uniformly alive; commits on main are. The same ten numbers answer 200 in objectstack, each an unrelated subject, so a repo-qualified re-cite would send a reader to the wrong record.
• No lychee config or workflow change, and no changeset: content/** is not a released package (check-changeset-presence below).

H1 reproduction, at the base 704e05b09d (origin/main at claim time)

• Lychee is not preinstalled in this container; the 0.18.1 release binary was fetched into the scratchpad. lychee --verbose --no-progress --config lychee.toml over the three files: 🔍 195 Total (in 15s) ✅ 182 OK 🚫 13 Errors, exit 2. Ten [404] lines, one per objectui issue URL, exactly the card's ten, on the same lines the card names (re-read at the base). The other three errors are this container's egress proxy, not link rot: github.com/changesets/changesets and github.com/lycheeverse/lychee answer 403 ("GitHub access to this repository is not enabled for this session") and git-cliff.org is a CONNECT 403. The CI run 36294593837 had 10 errors and none of these; they recur identically at the head.
• REST GET /repos/objectstack-ai/objectui/issues/N: all ten answer 404, on the API and on the HTML page. Lit controls: objectui#7013, docs: expressions/architecture guides teach text.value for expressions — switch all 23 occurrences to content, the ruled sole evaluation channel (objectstack#13670 option 2) #7015, feat(types,app-shell): carry the tool approval envelope through hydration #9229 and test(console,plugin-list): make two Storage.prototype spy instruments measure what they claim (objectui#7786) #7790 answer 200. Neighbours objectui#5792, fix(plugin-detail): derive ReferenceRailEntry from @objectstack/spec and retire the icon key #5794, chore(fields): narrow the datetime style cast, and make the #7443 changeset accurate for consumers #7792 and Studio: the "Create app" dialog's confirm control shares an accessible name with the toolbar button that opens it — a mis-click silently leaves the app uncreated #9231 also answer 404 and are cited nowhere in the swept trees.
• objectstack GET /repos/objectstack-ai/objectstack/issues/N: all ten answer 200 and every subject is unrelated (a spec docs generator PR, a bulkCreate card, a QA run, an email-template PR, and so on).

H2 and H3, site by site

Each row: the quoted anchor in the sentence, the dead citation, the replacement, and the evidence (the dead number's REST status is 404 in every row; the landing commit's subject is quoted without its conventional-commit prefix).

file	sentence anchor	old	new	evidence
ci-cd-pipeline.md	"three times so far, each found by a human and never by a gate (…, the third in the same package as the first)"	objectui#6943	"and a third in the same package as the first, all three recorded in 6aeba673c, the commit that landed this gate"	6aeba673c "require the directory form of the tooling exclude in every published build tsconfig" (PR 7409 answers 200). Its message records objectui#4006, #4836 and #6943 as the three incidents. #4006 and #4836 stay linked; both answer 200.
ci-cd-pipeline.md	"measured at two cards and three copied call sites (…), both closed by pointing the prose at resolveRecordSourceObjectName"	objectui#7627, objectui#7638	"measured at two cards and three copied call sites, both cards closed by pointing the prose at resolveRecordSourceObjectName (b041b9c0c, 2ce2612df)"	b041b9c0c "one shared record-source object-name reader, six plugins delegate" (PR 7637 answers 404); its message settles objectui#7627 by name. 2ce2612df "the record-page URL follows the record source, not the top-level key" (PR 7648 answers 200); abdcd189c's message names PR 7648 as the change that removed the spelling objectui#7638 was filed about, and its diff points the hook's @example at resolveRecordSourceObjectName (7 occurrences).
ci-cd-pipeline.md	"in #7638's card body and then in the dispatch that repeated it" (prose, not a link)	objectui#7638	"in the second card's body and then in the dispatch that repeated it"	The two cards are named in the preceding clause; the claim is unchanged.
ci-cd-pipeline.md	"a count ratchet would have been green too, which is why … rejected that option on the instance itself"	objectui#7753	"which is why 72498f257, the commit that landed this gate, rejected that option on the instance itself"	72498f257 "a census derives every handler key a registered renderer reads and requires its arm to declare it (objectui#7753)" (PR 7805 answers 404). Its message records the substitution that held RUNTIME_SLOT at 44 and ALL_SITES at 66, the count reading the sentence describes.
ci-cd-pipeline.md	"… then reproduced the identical defect on a THIRD writer neither guard covered, in that card's own required dogfood"	objectui#8057	"The identical defect was then reproduced on a THIRD writer neither guard covered, in that card's own required dogfood (recorded in 9662aca56, the commit that landed this gate)"	9662aca56 "apply the object-metadata write invariant at the write doors, not at an enumeration of writers" (PR 9238 answers 200). Its message: "objectui#8057 then reproduced the identical defect on a third writer in that card's own required dogfood, and a sweep found nine more". "That card" is objectui#7714, linked in the preceding sentence, which answers 200.
ci-cd-pipeline.md	"returns ZERO over the file #8057 is entirely about" (prose, not a link)	objectui#8057	"returns ZERO over the file that third writer lives in"	The third writer is referenced two sentences earlier; the claim is unchanged.
ci-cd-pipeline.md	"widened from content/docs/** alone by …"	objectui#7878	"widened from content/docs/** alone in 59a3a233d"	59a3a233d "widen the expression-carriage census to check:doc-types' surface (objectui#7878)" (PR 8105 answers 200).
ci-cd-pipeline.md	"cwd the only variable (…). Root AGENTS.md had taught"	objectui#7791	"cwd the only variable (223b1e4a3). Root AGENTS.md had taught"	223b1e4a3 "root the capability spec-parity pin at the file, not the cwd (objectui#7791)" (PR 7796 answers 200). Its message carries the two-cwd measurement.
ci-cd-pipeline.md	"The defect it closes (…):"	objectui#5793	"The defect it closes, repaired in the gate's own landing commit 111741454:"	111741454 "gate @objectstack/spec range floors against published symbols" (PR 6076 answers 200). Its message describes the plugin-detail ReferenceRailEntry defect in the sentence's own words, and its diff bumps packages/plugin-detail/package.json's @objectstack/spec floor from ^17.0.0 to ^17.1.0, so the same commit landed the gate and repaired the defect.
lookup.mdx	"and must never reach authored object metadata (…)"	objectui#7014	"and must never reach authored object metadata (0e3b3be09)"	0e3b3be09 "correct three false spec-alignment claims and pin the real boundary" (PR 7510 answers 200). e1545cfe6 (PR 7994), the commit that wrote this sentence, says it carries PR 7510's attribution. The claim itself is stale; see Acceptance note 1 for why the citation is kept resting on the landing commit rather than dropped.
select.mdx	"on both the editable and the read-only path (…)"	objectui#9230	"on both the editable and the read-only path (20b507aff)"	20b507aff "render a picklist option value when its label is blank (objectui#9230)" (PR 9258 answers 200).

All ten shas are unique at 9 characters (git rev-parse --short=9 returns 9 for each) and each is an ancestor of origin/main by construction (they were taken from git log origin/main).

H4 the whole tree, not only the ten

• git grep at origin/main over content/**, docs/**, README.md and skills/** for github.com/objectstack-ai/objectui/(issues|pull)/N: 191 distinct numbers across 14 files. REST on every one: 181 answer 200, 10 answer 404, and the ten are the card's ten. Nothing outside the three files.
• Lychee, same config plus --exclude '.*' --include for objectui issue and PR URLs (offline-safe: it reaches github.com only), over content/docs/**/*.md, content/docs/**/*.mdx, docs/**/*.md, docs/**/*.mdx, README.md and skills/**/*.md at the head 08be3d7708: 🔍 530 Total (in 14s) ✅ 188 OK 🚫 0 Errors 👻 342 Excluded, exit 0.
• Bare-number mentions of the same ten (objectui#N in prose, which lychee cannot see) elsewhere in the tree at origin/main: 204 sites outside the three files: 38 in packages/*/src non-test text (JSDoc and comments, some of which ship in dist per PR objectui#10707's dist reading), 20 in pending .changeset/*.md files (which publish verbatim into the CHANGELOG at the next release), 81 in test files, 28 in scripts/, 6 under .github/, 1 in governed agent text, 30 elsewhere. Not widened into this PR; listed under Acceptance notes for a card.

Gates, all on the head 08be3d7708

• Lychee over the three files, repo config: 🔍 185 Total (in 17s) ✅ 182 OK 🚫 3 Errors, exit 2 — the 3 are the container proxy artefacts named under H1 and zero are objectui links (the base run had 13 errors, 10 of them these links). Objectui-scoped (--exclude '.*' --include objectui issue/PR URLs): 🔍 185 Total (in 13s) ✅ 173 OK 🚫 0 Errors 👻 12 Excluded, exit 0.
• pnpm docs:check-links (check-doc-links): Links are valid across 17 scan roots., exit 0.
• pnpm check:control-bytes: check-control-bytes: OK (scanned 8943 tracked text file(s); skipped 85 binary)., exit 0.
• pnpm check:new-line-citations: VERDICT new-cross-file-line-citations: 0 new citation(s), enforcement report-only -> exit 0.
• node scripts/check-changeset-presence.mjs: 3 file(s) changed, 0 of them published source of a package the release covers … No source or published contract of a released package changed in this range, so no changeset is owed., exit 0.
• pnpm check:doc-fences exit 0; pnpm check:doc-example-ids exit 0 (414 real reference(s) all resolve in the catalog registry).
• The test readers node scripts/markdown-test-inputs.mjs --list names for the classes content/docs/** and content/docs/guide/ci-cd-pipeline.md (26 files, including ci-cd-pipeline-doc.test.ts and check-links-workflow.test.ts), run as one pnpm exec vitest run from the repo root: Test Files 26 passed (26), Tests 1195 passed (1195).
• Frontmatter lines 1 to 4, md5 before and after, identical: ci-cd-pipeline.md 6ca230b6b64dac22a2031d21ff311b13, lookup.mdx 289a5c887fd231f6b0b0824f53769fad, select.mdx bae97ed7e436e0a06b8a154242ec15f1.
• NOT MEASURED: the Check Links workflow itself, which runs on schedule and workflow_dispatch only. Per the triage, re-run it via workflow_dispatch once this lands to show the weekly sweep green.

Serial

• At claim time (06:07Z) no open PR touched the three files. Before the final push, origin/main was fetched into the private ref refs/issue-10755/main = f308a655b8, two commits past the base (f308a655b8, baac95a261). git diff --stat between the base and that ref over the three files is empty, and git merge-tree --write-tree HEAD refs/issue-10755/main exits 0, so no merge commit is owed.

Acceptance notes

1. The lookup.mdx paragraph around the re-cited sentence is a stale contract claim (not widened; for a card). It says the installed spec's SelectOptionSchema "is strict over exactly {label, value, color, default, visibleWhen} and refuses description by name", that authoring it "fails the whole field with a 422", and that the key "must never reach authored object metadata". Measured on the installed @objectstack/spec 17.4.0 (@objectstack/spec/data): SelectOptionSchema.safeParse of an option with description is ACCEPTED, FieldSchema with that option is ACCEPTED, and the refusal control icon is REFUSED with unrecognized_keys. Spec 17.3.0 declared the key under the maintainer's 2026-08-25 ruling on objectui#6140 / objectui#6153 (Option A); 544ecba84f (PR 9588) already moved the shipped docblocks off the refusal, and packages/types/src/__tests__/select-option-spec-extension-7014.test.ts pins the moved boundary. The repair is a paragraph rewrite, outside this order's surface (citations only). The citation at that site is kept resting on the landing commit 0e3b3be09, the commit that made the measurement the sentence still repeats, so a reader following it lands on when, and against which spec version, the claim was true; dropping the pointer would leave the stale sentence with nothing to date it by.
2. Sibling class, one sweep card rather than ten: the 204 bare-number mentions of the same ten numbers outside the three files (H4, last bullet). The shipping subset is the 38 packages/*/src non-test sites and the 20 pending changesets; objectui#7014 alone has 30 mentions, including three pending changesets named after it. This is the class objectui#10701 was filed for, and that card itself suggested one sweep card if more numbers in the range were gone.
3. Neighbouring numbers that also answer 404 (objectui#5792, fix(plugin-detail): derive ReferenceRailEntry from @objectstack/spec and retire the icon key #5794, chore(fields): narrow the datetime style cast, and make the #7443 changeset accurate for consumers #7792, Studio: the "Create app" dialog's confirm control shares an accessible name with the toolbar button that opens it — a mis-click silently leaves the app uncreated #9231) are cited nowhere in the swept trees; observation only.
4. Instrument versions: local lychee is 0.18.1; lycheeverse/lychee-action@v2 runs the current release. Same lychee.toml, same globs; the CI reading is the authoritative one, and the three residual errors above are container-specific.

---

Generated by Claude Code

[github-actions]

changeset-claim-re-read

⚠️ 1 pending changeset(s) describe a file this change touches

Their bodies publish verbatim into the CHANGELOG at the next release, so this is a request to re-read them against your diff — addressed here because you are the one seat that can answer it without re-deriving anything.

⛔ Nothing here blocks, and nothing here is a verdict on your change. This gate exits 0, is not a required context, and judges name resolution, never meaning: it asked whether a pending body names a file you touched. "Is this sentence still true?" is the one question it will not answer, and the one you are being asked to answer.

.changeset/7155-converge-lookup-dialect-camelcase.md

• names content/docs/fields/lookup.mdx → content/docs/fields/lookup.mdx — edited by this change

  Also moved with the rename: content/docs/fields/lookup.mdx and user.mdx (whose snippets CI compiles against the built d.ts), all seven in-repo producers, and the inline-edit enrichment allow-list in @object-ui/plugin-detail. plugin-grid's relationalMetaKeys.ts drops the four legacy-alias verdicts and retires that verdict class; its gate is restated to assert the class no longer exists rather than passing vacuously.

Read the paragraph, not the line: both false halves of the objectui#8617 claim sat in one paragraph, and correcting either alone would have left it asserting the same wrong thing.

If a claim did go false, correct the body. That is precedented and prose-only, frontmatter untouched; check-changeset-overwrite.mjs will report the correction as its own case 2 ("correcting a declaration on purpose … legitimate"), which is the intended shape — one gate asks for the read, the other records the write.

Not covered, stated so nobody reads this as more: a born-false claim that spells no line address at all (objectui#9495 coordinated one by ORDINAL — "a grep finds that member first" — and deciding that means reading what the sentence means), a claim spelled as a symbol or a package rather than a backticked file name, and a file named ambiguously.

Compared the checked-out tree with f308a655b (merge-base with origin/main): 3 file(s) changed outside .changeset/, read against 1575 pending declaration(s) that publish a body (2163 pending in total). · run

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 08be3d77080e722d9fff9524b947fabc43630c05

① Derived judgments

Premise re-measured. REST /issues/N for the ten (5793, 6943, 7627, 7638, 7753, 8057, 7878, 7791, 7014, 9230): all 404; HTML pages also 404 (7014 probed). Controls 7013, 7015, 9229, 7790 answer 200. Landing PRs 7637 and 7805 answer 404; 7409, 7648, 9238, 8105, 7796, 6076, 7510, 9258 answer 200. So the "commits, not PRs" choice is forced, not stylistic.

The ten commits. Every sha resolves, is an ancestor of refs/review/main (25c7d58) by merge-base --is-ancestor, is unique at 9 chars (rev-parse --short=9 returns 9 for each), and answers 200 both at /repos/.../commits/sha and at the HTML /commit/sha page. Per site, subject and diff read:

• 6943 → 6aeba673c (PR 7409): adds check-published-tsconfig-tooling-exclude.mjs; its message names objectui#4006, finding: @object-ui/core 的 build 程序把 __benchmarks__ 一起编译,core.bench.js/.d.ts 落进已发布的 dist/(#4006 同族,新实例) #4836, check:published-dist is RED on main: @object-ui/fields ships dist/__tests__/numberInputBrowserReadings.d.ts in its tarball #6943 as the three published-dist incidents. "all three recorded in 6aeba673c, the commit that landed this gate" is true. @object-ui/fields 与 plugin-editor 的 build 程序把测试文件一起编译,73 个 *.test.d.ts 落进已发布的 dist/ #4006 and finding: @object-ui/core 的 build 程序把 __benchmarks__ 一起编译,core.bench.js/.d.ts 落进已发布的 dist/(#4006 同族,新实例) #4836 stay linked and answer 200.
• 7627 → b041b9c0c (PR 7637): "Fixes finding(core): the six per-block data.provider === 'object' folds could collapse onto one shared reader — the seventh now lives in normalizeListViewSchema #7627"; publishes resolveRecordSourceObjectName, useSettledSchema docblock stops prescribing the ladder.
• 7638 → 2ce2612df (PR 7648): message does not name 7638, but the diff carries .changeset/7638-navigation-url-follows-record-source.md and three *-7638.test.* files, and rewrites the hook's @example to resolveRecordSourceObjectName (7 occurrences). It is 7638's landing commit; "both cards closed by pointing the prose at resolveRecordSourceObjectName (b041b9c0c, 2ce2612df)" is true of both.
• 7753 → 72498f257 (PR 7805): subject names objectui#7753; adds check-handler-key-read-sites.mjs, edits this very doc row; its message records the SUBSTITUTION holding RUNTIME_SLOT at 44 and ALL_SITES at 66, which is the count-ratchet-would-be-green reasoning the sentence describes. Attribution moves from the card to the commit whose message records the same reasoning; the claim itself is unchanged.
• 8057 → 9662aca56 (PR 9238): message: "objectui#8057 then reproduced the identical defect on a third writer in that card's own required dogfood, and a sweep found nine more"; adds check-object-metadata-write-doors.mjs, so "the commit that landed this gate" is true. "That card" is Designer PUTs half-filled lookup drafts that @objectstack/spec 17.3.0 refuses (reference is now required) — keep the draft client-side and do not PUT it (#7122 ruling, item 4 = A) #7714, linked in the preceding sentence, 200.
• 7878 → 59a3a233d (PR 8105): subject names objectui#7878; widens the census, edits this doc line.
• 7791 → 223b1e4a3 (PR 7796): subject names objectui#7791; the diff's test-file header carries "7 passed from the repo root, 2 failed / 5 passed from the package", the sentence's exact figures.
• 5793 → 111741454 (PR 6076): .changeset/5793-spec-range-floors.md, adds check-spec-range-floors.mjs and its workflow, and bumps packages/plugin-detail/package.json @objectstack/spec ^17.0.0 to ^17.1.0. "repaired in the gate's own landing commit" is true.
• 7014 → 0e3b3be09 (PR 7510): .changeset/7014-select-option-spec-claim-correction.md, so it is 7014's landing commit. git blame at base attributes all four lines of the sentence to e1545cfe6 (PR 7994, settles docs(fields): lookup.mdx still claims option description is aligned with @objectstack/spec — the spec refuses it by name #7537), whose message says the page "carries the same attribution PR fix(types,fields): correct three false spec-alignment claims and pin the real boundary #7510 gave". Claim unchanged and true of that commit's measurement (spec 17.2.0). The claim is false of the installed spec today; see ③.
• 9230 → 20b507aff (PR 9258): subject names objectui#9230; wrote this select.mdx section (+16 lines).

Two prose re-points. "in the second card's body" follows "measured at two cards and three copied call sites" in the same sentence, so "the second card" resolves. "the file that third writer lives in" refers back to "a THIRD writer neither guard covered" one sentence earlier; it is referenced, not named by file (the PR body's "named two sentences earlier" overstates), and the old wording named no file either. Both claims unchanged.

Nothing lychee-visible still 404s. The three head files carry 171 distinct objectui issue/pull links; the swept tree (content/**, docs/**, README.md, skills/**) carries 181 distinct numbers in 14 files (main: 191 in 14; the difference is the ten). REST on all 181: 181 answer 200. No bare #N prose mention of the ten remains in the three files. Lychee 0.18.1 (the dev's binary), repo lychee.toml, three head files exported to scratch: 185 total, 182 OK, 3 errors, exit 2; the 3 are git-cliff.org (proxy TLS), github.com/lycheeverse/lychee and github.com/changesets/changesets (proxy 403), none an objectui link. Objectui-scoped: 185 total, 173 OK, 0 errors, 12 excluded, exit 0. Tree-wide objectui-scoped: 530 total, 188 OK, 0 errors, 342 excluded, exit 0. All match the PR body's figures. Note the new citations are bare backticked shas, which lychee never reads (and lychee.toml excludes /commit/ URLs anyway): the sweep proves the dead links are gone; the new pointers' liveness rests on ancestry of main, which holds.

Published text. Word-level diff shows only the citation spans and the two re-points move; every changed sentence judged above. Frontmatter md5 (lines 1 to 4 and the whole block) identical base to head: 6ca230b6b64dac22a2031d21ff311b13, 289a5c887fd231f6b0b0824f53769fad, bae97ed7e436e0a06b8a154242ec15f1. PR body spot checks: "five of six existing sha citations are 9 chars" true (71be244d52 is the 10); "+6/−6, 3 files" true; H4 "191 numbers, 14 files" true; abdcd189c line 29 does name PR 7648 as removing the spelling 7638 was filed about; serial claims hold (below).

② Semver level

None. content/** only. check-changeset-presence.mjs isPublishedSource counts a path only as pkg/src/, pkg/index.html, or a files-listed entry of a fixed-group package; content/ is under no package and @object-ui/site is in ignore. No changeset owed; "Changeset Declaration" is green on the head.

③ Boundary flags

• CI on head: 39 check runs, 36 success, 3 skipped by design (Test (coverage), coverage shard, dependabot). check-links runs on schedule and workflow_dispatch only (.github/workflows/check-links.yml), so it is not among them and no PR can show the red-then-green; a workflow_dispatch after merge is the only instrument.
• Merge: git merge-tree --write-tree refs/review/pr-10766 refs/review/main (main now 25c7d58, four commits past base 704e05b) exits 0; the three files are untouched on main since base.
• Open PRs: 9 others open; none touches the three files.
• Governed: check-governed-queue-guard.mjs --test on the three paths: NOT GOVERNED (0 of 5 surfaces).
• Draft and assignee: draft is correct: AGENTS.md exception ② holds a dispatched PR at draft for the dispatch seat to flip ready and queue. PR has no assignee; the card has os-elon-musk. os-dev.md rule 2 says the dev never writes the assignee, so the refused write was outside the dev's contract and its refusal costs nothing; the PR assignee is the dispatch seat's field.
• Out-of-scope finding, confirmed by my own measurement on the installed @objectstack/spec 17.4.0 (/home/user/objectui-issue-10712/node_modules, ./data export): SelectOptionSchema keys are color, default, description, label, value, visibleWhen; an option with description is ACCEPTED; a select FieldSchema carrying it is ACCEPTED; the icon control is REFUSED with unrecognized_keys. So lookup.mdx lines 68 to 76 ("strict over exactly {label, value, color, default, visibleWhen}", "refuses description by name", "fails the whole field with a 422", "must never reach authored object metadata") misstate the installed schema. Pre-existing: 544ecba84f (PR 9588, 2026-09-16) corrected the five shipped docblocks and left this page, and its card When the @objectstack/spec pin moves to 17.3: PR #7510 / #7520's comment "repairs" become FALSE — rows and options[].description are now declared keys #7635 now answers 404. It is a paragraph rewrite, outside a citations-only order, so follow-up, not this PR; but no card exists yet (147 open issues scanned; only finding(docs): the weekly check-links run is red on main — ten docs links cite objectui issues that answer 404 (ci-cd-pipeline.md ×8, lookup.mdx, select.mdx) #10755 mentions lookup.mdx), so the dispatch seat owes one before accepting. The kept citation is defensible under the triage ("cite the landing commit"), and it now dates the claim to the 17.2.0 measurement, which is the most a citation can do for a sentence that says "installed".
• Nit, not blocking: the PR body's "The third writer is named two sentences earlier" should read "referenced two sentences earlier"; the doc never names the file.

Implemented-by: claude/issue-10755-dead-issue-citations
Reviewed-by: session_014mXUNuFomfj24w7s1pZzhN

VERDICT: PASS