docs(ci-cd-pipeline): state the link-sweep populations instead of copying their counts (objectui#7886)

[claude]

Fixes #7886

content/docs/guide/ci-cd-pipeline.md's Link Checking (check-links.yml) section published the same two hand-copied document counts that objectui#7825 had just removed from that workflow's own header — in the present tense, on a page the site serves to readers, which is the most exposed of the four instances of the objectui#7448 class.

Measured on origin/main (d9580f464) at the time of this change:

claim	page said	measured today
content/docs/** markdown	183	184
repo-root docs/** markdown	15	17

find content/docs -type f \( -name '*.md' -o -name '*.mdx' \) | wc -l   # 184
find docs -type f \( -name '*.md' -o -name '*.mdx' \) | wc -l           # 17

check:shell-escape-residue's own verdict line corroborates the first reading independently, from a separate walk of the same tree: content/docs: 184 file(s), 1064 fence(s).

Not a number refresh

Changing 183 to 184 and 15 to 17 would only reload the trap — nothing reads either figure, so nothing would go red the next time they drift. This applies the remedy this repository has now landed three times (objectui#7448 twice, objectui#7825 once), and that this very page already applies to its own workflow count at the top ("This page deliberately states no workflow count"): state the population, and point at the reading.

The paragraph now names which trees, which extensions, and the apps/site/source.config.ts declaration the content tree comes from (dir: '../../content/docs', baseUrl /docs), says that the args glob list decides the scope and nothing else, and points at the two live readings — a find in any checkout, and Lychee's own run summary. No numeral survives in the section that adding or deleting a document could falsify: the only digits left are issue references (objectui#7448, objectui#7825, objectui#7886, objectui#3449), which the class's own pin excludes by construction.

Historical vs present tense

Read as the page actually worded it rather than assumed from the workflow header. The count-bearing sentence was wholly present tense ("Lychee sweeps both documentation trees…"); the historical claim was a separate sentence, already past tense, already carrying its PR number (objectui#3449), already carrying no live numeral — the shape the ruling asks for. It is kept, split into its own paragraph and labelled "Past tense on purpose — none of the following describes the scope today", so the boundary between the live claim and the history is visible rather than inferred. The still-live half of that sentence (the test derives the expected scope from the site config) moved up into the live paragraph, where it belongs.

objectui#3653's job table did not move — before and after, both green

This change adds and removes no ci command. The prose edited is a paragraph below the two-link-checker comparison table and unrelated to the run:-step job table objectui#3653 pins; confirmed by reading the section before editing, and by scripts/__tests__/ci-cd-pipeline-doc.test.ts carrying no mention of Lychee, check-links, sweeps or either count.

• Before the edit: Test Files 1 passed (1) · Tests 36 passed (36) (os-verify-lock: VERDICT command-exit 0)
• After, on the committed SHA: Test Files 1 passed (1) · Tests 36 passed (36) (os-verify-lock: VERDICT command-exit 0)

PR #7868's docs row for the report-only census step is untouched; the diff is 21 insertions and 7 deletions, all inside the one paragraph.

Measured: no pin watches these two sentences

The card asked for this reading, and explicitly not for the pin itself.

The shape is exactly the class. Run the pin regex objectui#7825 landed (POPULATION_COUNT in scripts/__tests__/check-links-workflow.test.ts) over the pre-fix paragraph and it fires on both counts — ["183 pages", "15 files"]. The narrower objectui#7448 twin regex (check-doc-fence-languages.test.ts, check-doc-component-types.test.ts) fires on both as well. Over the rewritten paragraph both return [].

Nothing was reading it. All three copies of the pin read .github/workflows/*.yml headers (the third also reads its own test file's header); none reads content/docs/**. Nine other test files read this page — check-doc-expression-carriage, check-doc-example-shared-reader, check-handler-key-read-sites, check-skill-examples, check-skill-eval-tokens, merge-queue-reporting, lint-workflow, doc-version-claims, ci-cd-pipeline-doc — and not one asserts anything about the Lychee prose or either figure. A repo-wide grep for the two literals finds them only here and, as deliberate historical fixtures, inside check-links-workflow.test.ts.

So the gap is real and unguarded, and it is reported rather than built: a fourth copy would introduce a new scan population over the content/docs tree, which this lane requires to land report-only first, and objectui#7888 is already working that pin family's own narrowness problem. Filed as an out-of-scope finding rather than carried here.

Verification

Gate union re-run after the final commit, on d6bf5f79a, with git diff HEAD empty (0 bytes):

gate	exit	verdict
scripts/__tests__/ci-cd-pipeline-doc.test.ts + 11 other readers of this page	0	Test Files 12 passed (12) · Tests 372 passed (372)
check:doc-types	0	✅ Every documented component type is registered.
check:doc-snippets	0	Every covered documentation snippet compiles against the built types. (after the scoped closure build the gate's own --build-filter names — its first run was EXIT=2, precondition not met, NOT MEASURED)
check:doc-fences	0	✅ check:doc-fences — every TypeScript block in 227 document(s) is fenced ts/tsx/typescript…
check:shell-escape-residue	0	✅ check-shell-escape-residue: OK (5/5 root(s) resolved …)
docs:check-links	0	Links are valid across 17 scan roots.
check:control-bytes	0	✅ check-control-bytes: OK (scanned 6393 tracked text file(s); skipped 85 binary).
check:doc-example-readers	0	OK 80 documented symbol(s) … no @example hand-spells one.
check:handler-key-reads	0	OK 106 arm(s) … every judged read is a declared member of its arm.

Governed surface, verbatim: ✅ NOT GOVERNED — 1 path(s) checked against 5 governed surface(s); none matched. / An ordinary pull request: the normal review and merge-queue route applies.

Changeset, verbatim and obeyed rather than reasoned about: Compared the working tree with d9580f464 (merge-base with origin/main): 1 file(s) changed, 0 of them published source of a package the release covers, 0 of them a manifest whose published contract moved, 0 under a package changesets ignores, 0 changeset(s) added. / ✅ No source or published contract of a released package changed in this range, so no changeset is owed. — so no changeset is added, not even the empty-frontmatter form.

Repo-wide lint, narrowed and the narrowing measured (three readings, not an assumption):

1. Population read from eslint's own config, not guessed: every files: block in eslint.config.js targets **/*.{ts,tsx}; the single md occurrence in that file is inside a comment.
2. File count read from --format json: eslint over the changed path returns one entry, "File ignored because no matching configuration was supplied.", 0 errors — the file is not in the judged population at all.
3. Invariance for untouched files: the diff modifies one .md file and adds or removes no .ts/.tsx, so eslint's judged population is byte-identical before and after and no untouched file's verdict can move.

Clause-② is no: prose only, no gate predicate and no scan population changed.

---

🤖 Generated with Claude Code

Generated by Claude Code

---

Generated by Claude Code

[claude]

The out-of-scope finding this PR reports under Measured: no pin watches these two sentences is filed as objectui#7901 — no assignee, label finding, left for triage. It is the coverage half of the family (which surfaces the pin reads at all); objectui#7888 is the narrowness half (the numeral and the noun must be adjacent). Separable, and the issue records why doing objectui#7888 first makes the census objectui#7901 needs honest on its first run.

No pin was built here, per the card.

---

Generated by Claude Code