You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
finding(gate): scripts/check-doc-fence-languages.mjs's header still says its scan surface is check-doc-snippet-types' "exactly" and that the coupling pin fails on any differing document list — both stopped being true when PR #8158 added the docs/*.md leg to the snippet gate only #8162
Filed by the domain:devx @ objectui execution seat (PM session session_01FhBNJcLRZLe8M87VcUgpKr, R46, 2026-09-06T21:38Z), from the report on objectui#7856 card 1 (PR #8158, head f011733bf). The dev flagged it as owed follow-up outside that card's file surface (the surface named the gate's TEST, not the gate), and recommended a one-paragraph follow-up card rather than widening a landed PR's surface after the fact. The seat agrees.
What is stale
scripts/check-doc-fence-languages.mjs's header (on f011733bf, unchanged from 571b4870d) states:
"The scan surface is check-doc-snippet-types's, exactly: …" — no longer exact. PR feat(gate): bring the root docs/*.md pages into the doc-snippet walk #8158 gave the snippet gate a docs/*.md leg (ROOT_DOCS / rootDocsPages()); check:doc-fences did not gain it, by design (its surface is not objectui#7856 card 1's to move).
The coupling pin "fails if they ever return different document lists" — it no longer asserts plain equality. scripts/__tests__/check-doc-fence-languages.test.ts's walks exactly the documents the snippet gate walks now subtracts exactly rootDocsPages() from the snippet gate's walk before comparing, and pins that the subtraction is non-empty and that the leg stops above docs/adr/ and docs/audits/ in both walks.
Why prose, not enforcement
Nothing is unguarded: the pin itself carries the accurate statement and fails on any other drift. The header is what a reader (or an agent) reads first, and it now promises a stronger equality than the pin holds. The fix is a one-paragraph header edit on that gate naming the divergence and its reason (objectui#7856 card 1 / card 2 boundary), and no test change.
Verification recipe
git fetch origin main
git show origin/main:scripts/check-doc-fence-languages.mjs | grep -n -A3 "scan surface"
git show origin/main:scripts/__tests__/check-doc-fence-languages.test.ts | grep -n "rootDocsPages"
Routing note: this could ride with objectui#7856 card 2 (the governed docs/adr/** half, another seat's PR) but should not — it is an ungoverned one-header edit and card 2 stops in draft for a human, which would delay this correction for no benefit.
Filed by the
domain:devx @ objectuiexecution seat (PM sessionsession_01FhBNJcLRZLe8M87VcUgpKr, R46, 2026-09-06T21:38Z), from the report on objectui#7856 card 1 (PR #8158, headf011733bf). The dev flagged it as owed follow-up outside that card's file surface (the surface named the gate's TEST, not the gate), and recommended a one-paragraph follow-up card rather than widening a landed PR's surface after the fact. The seat agrees.What is stale
scripts/check-doc-fence-languages.mjs's header (onf011733bf, unchanged from571b4870d) states:check-doc-snippet-types's, exactly: …" — no longer exact. PR feat(gate): bring the root docs/*.md pages into the doc-snippet walk #8158 gave the snippet gate adocs/*.mdleg (ROOT_DOCS/rootDocsPages());check:doc-fencesdid not gain it, by design (its surface is not objectui#7856 card 1's to move).scripts/__tests__/check-doc-fence-languages.test.ts'swalks exactly the documents the snippet gate walksnow subtracts exactlyrootDocsPages()from the snippet gate's walk before comparing, and pins that the subtraction is non-empty and that the leg stops abovedocs/adr/anddocs/audits/in both walks.Why prose, not enforcement
Nothing is unguarded: the pin itself carries the accurate statement and fails on any other drift. The header is what a reader (or an agent) reads first, and it now promises a stronger equality than the pin holds. The fix is a one-paragraph header edit on that gate naming the divergence and its reason (objectui#7856 card 1 / card 2 boundary), and no test change.
Verification recipe
Routing note: this could ride with objectui#7856 card 2 (the governed
docs/adr/**half, another seat's PR) but should not — it is an ungoverned one-header edit and card 2 stops in draft for a human, which would delay this correction for no benefit.Refs objectui#7856, PR #8158.