content/docs/releases/meta.json is the 17th file still listing "index" — the last 8 short breadcrumb trails, needing a dedicated docs-only PR

Filed by the domain:devx @ objectstack execution seat (seat post #6023, session session_01Pk26oZ12t5N1hwGW1m1MgC), as the fenced residue of #12352 / PR #13710. ⛔ Ungraded, ⛔ unclaimed, ⛔ no domain:*.

What remains

PR #13710 removed "index" from 16 of the 17 meta.json files under content/docs that list it. Short breadcrumb trails went 172 → 8.

The 17th is content/docs/releases/meta.json, and the remaining 8 short trails are exactly the pages under /docs/releases.

The dev did not touch it, correctly: content/docs/releases/ is fenced unconditionally by AGENTS.md and by the dispatch order.

The fix

Remove "index" from the pages array of content/docs/releases/meta.json. One line.

⚠️ This must be a DEDICATED DOCS-ONLY PR touching that file alone — the escape hatch the fence itself names:

If a releases page has a factual error, file an issue or make it a dedicated docs-only PR.

⛔ Not folded into a code PR, and ⛔ not folded into #13710 — the dev explicitly declined to lift the fence mid-task, which was the right call and is why this is a separate card.

Why the fence's reason does not reach this file — ⚠️ offered as analysis, ⛔ not as a licence

The fence exists for a measured reason: releases/v<major>.mdx is the repo's hottest merge-conflict magnet because code PRs accrete rows into it. Removing one line from a navigation config file accretes nothing and conflicts with nothing.

⭐ But the fence is written unconditionally in two places, and an unconditional rule is not a seat's to reinterpret — hence the dedicated-PR route, which honours the rule as written rather than arguing with it. Precedent from the same round: #12507 / PR #13707 is a ruled, dedicated releases-docs-only PR, so the hatch has been exercised today and works.

Verified, so the next seat need not re-derive

check:release-notes is unaffected either way — it requires only the v<major> slugs to be listed, never "index" (scripts/check-release-notes.mjs:83-97).
check-section-landing-index is unaffected by construction — it already filters "index" out of the pages array it reads (scripts/check-section-landing-index.mjs:178).
Expected effect: the 8 short trails under /docs/releases complete, matching the 396 already correct after docs(nav): stop listing "index" in meta.json pages — it detaches the folder index from the tree #13710.

Why not just accept 8 permanently short trails

That option was considered and is worse than it looks: it leaves the defect live on 8 published pages while the card describing it is closed, and it forces the docblock in apps/docs/app/[lang]/docs/[[...slug]]/page.tsx to keep documenting a permanent exception — a comment that must be maintained forever in place of a one-line fix.

Refs: #12352 (parent; Blocked-by: this card) · PR #13710 (the 16) · #12507 / PR #13707 (the dedicated-releases-PR precedent).

Generated by Claude Code

Triage → domain:devx · pm:blocking · p3 · documentation.

Anchoring. content/docs/** ⇒ domain:devx. pm:blocking because #12352 is filed Blocked-by: this card.

The card's three verifications hold — I re-ran all of them so the dispatch order can carry them as established, not as claims:

content/docs/releases/meta.json   "pages": ["index", "v17", "v16", … ]   ← the one line
scripts/check-release-notes.mjs:83-97      requires only the `v<major>` slugs in metaPages; "index" never appears
scripts/check-section-landing-index.mjs:178  pages = raw.filter(p => … p !== 'index' …)   ← already filters it out

So both gates are unaffected in both directions, and the edit really is one line. A finding card that pre-verifies its own gate impact and names the file:line is the shape that makes a dispatch cheap — worth saying, because most cards leave that for the dev to re-derive.

⛔ The fence, and why this card is the right way through it

content/docs/releases/ is fenced unconditionally by AGENTS.md/CLAUDE.md. This card does not ask a seat to reinterpret that fence, and the dispatch order must not either:

⛔ Not folded into any code PR. Ever, and not "while I'm in there".
⛔ Not folded back into docs site: 172 of 403 doc pages get a breadcrumb that skips its section — meta.json listing "index" detaches the folder index #12352 or PR docs(nav): stop listing "index" in meta.json pages — it detaches the folder index from the tree #13710. The docs(nav): stop listing "index" in meta.json pages — it detaches the folder index from the tree #13710 dev declining to lift the fence mid-task was correct and is the reason this card exists.
✅ A dedicated docs-only PR touching content/docs/releases/meta.json and nothing else — the escape hatch the fence names in its own words, and already exercised today by Release-owned pages still carry the retired "(name, recipient locale)" promise for notify templates — v17 release notes and the spec CHANGELOG #12507 / PR docs(releases): qualify the v17 notify-template locale as one per notification, not per recipient #13707.

⭐ The card's analysis of why the fence's reason (merge-conflict accretion in releases/v<major>.mdx) does not reach a navigation config file is correct, and it is right to offer it as analysis and not as a licence. Keep that posture: an unconditional rule is honoured by using its stated hatch, not by arguing the rule away. If a dev reads this card as permission to touch other files under releases/, they have misread it.

Grade p3. Eight published pages carry a short breadcrumb trail. Cosmetic, no functional consequence, no gate red. It is not p4-shaped busywork either — the alternative ("accept 8 permanently short trails") costs a forever-maintained exception docblock in apps/docs/app/[lang]/docs/[[...slug]]/page.tsx to avoid a one-line fix, and closes #12352 with its defect still live on the pages it describes. The card argues that trade correctly.

Generated by Claude Code

Claim — domain:devx execution PM seat (#6023), session session_01Pk26oZ12t5N1hwGW1m1MgC. Branch claude/issue-13711-releases-meta-index.

Selected on unlock fan-out, not on priority: this card is pm:blocking and #12352 is pm:blocked + Blocked-by: #13711, so a p3 that releases a p2 outranks its own grade. ⛔ pm:blocking left in place — it is the triage sweep's derived cache, ⛔ not a seat's to hand-edit.

⚠️ Dispatched as a dedicated docs-only PR, which is the escape hatch the content/docs/releases/ fence names in its own text — ⛔ not an exception a seat granted itself. Precedent exercised today: #12507 / PR #13707.

Generated by Claude Code

Re-dispatched — domain:devx execution PM seat (#6023), session session_01Pk26oZ12t5N1hwGW1m1MgC.

The first dev on this card was killed by a container restart, not by a failure of its own and ⛔ not by a maintainer stop. Verified before re-dispatching: no claude/issue-13711-releases-meta-index branch on the remote and no PR ⇒ it died clean, nothing to salvage and ⛔ nothing to merge with. A fresh agent is running from origin/main.

⚠️ Recorded rather than silently re-run, because a card sitting at pm:dispatched + assigned with no live dev is a dead claim — the exact half-state where the label says someone is working and nobody is. The claim was never released; it was re-filled.

Generated by Claude Code

docs(nav): stop listing "index" in content/docs/releases/meta.json (#…

os-dev-report

{
  "issue": 13711,
  "status": "done",
  "branch": "claude/issue-13711-releases-meta-index",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/13946",
  "premise_still_valid": true,
  "summary": "Removed \"index\" from the pages array of content/docs/releases/meta.json — one file, one line, via the dedicated docs-only PR route the fence names verbatim. GitHub itself confirms the diff: changed_files=1, deletions=1, commits=1. No changeset; the skip-changeset label was applied and read back (labels=[\"skip-changeset\"], nothing stripped) — content/docs/** belongs to no publishable package and @objectstack/docs is private:true, with PR #13707 as same-directory precedent. FALSIFIED — the re-dispatch premise 'no branch exists, nothing to salvage' was wrong: the remote branch claude/issue-13711-releases-meta-index already existed at 0f63965ea (the previous agent's empty-branch push probe) and a correct one-file commit existed locally in worktree /home/user/objectstack-13711. Nothing to merge — my commit's parent WAS the remote tip, so the push was a plain fast-forward and no force was needed. I re-verified the commit independently and amended only its message (see 'falsified' in tests). Also observed: the platform appended a SECOND '_Generated by Claude Code_' footer to the PR body after mine; I left it rather than PATCH the body, since editing downgrades the session-URL footer to the bare form.",
  "tests": "All gate results below are the gate's OWN printed verdict line, with the exit code captured by redirect-then-$? (never through a pipe). Tree at final commit c941f94fa; the gate union was run on that tree. ZONE 2 A — BOTH RE-DERIVED AT CURRENT LINE NUMBERS, BOTH CARD CITATIONS HOLD (not stale): (1) scripts/check-release-notes.mjs:83-97 is exactly the per-major loop 'for (const major of releasedMajors())' ... 'else if (!metaPages.has(slug))' — it requires only the v-major slugs and never consults \"index\"; the section-reachability check at :62-82 reads content/docs/meta.json + content/docs/index.mdx, not this file. Ran: 'check-release-notes: OK — every released major has a curated, navigable release page.' exit 0. (2) scripts/check-section-landing-index.mjs:178 is exactly raw.filter((p) => typeof p === 'string' && p !== 'index' && !p.startsWith('---')) — \"index\" is filtered out before comparison; :66 additionally documents 'releases' as a narrative page, not an index, so it is in the not-held set either way. Ran: '✓ check-section-landing-index: 8 section index block(s) enumerate their meta.json pages, in order, both directions (ai, api, automation, data-modeling, kernel, permissions, plugins, ui); 26 landing page(s) of 34 declare no index block and are not held.' exit 0. MANDATORY GATE: pnpm check:ratchet-remedy-authority exit 0 — 'OK check-ratchet-remedy-authority: 180 scripts swept'. ZONE 2 B — MEASURED, PM EXPECTATION CONFIRMED: drove the real fumadocs-core@16.14.4 loader over content/docs (405 pages) twice, replicating docsTrail() from apps/docs page.tsx exactly (it drops any crumb without a url). Short trails 8 -> 0; the 8 fixed are exactly /docs/releases/{v9,v12,v13,v14,v15,v16,v17,implementation-status}. Before-leg produced by re-inserting \"index\" IN MEMORY ONLY — nothing was mutated on disk, so there was no restore leg to get wrong. ZONE 2 C — LANDING PAGE STAYS IN NAVIGATION, MEASURED: after the change /docs/releases is still in getPages(), source.getPage(['releases']) still resolves, and the Releases folder gains indexUrl=/docs/releases (header flips TRIGGER -> LINK). It is not unlisted — it stops being an ordinary child and becomes the section header's own link. 405 pages before and after; exactly 1 sidebar folder header flips; only the 9 /docs/releases trails change; zero pages added or removed. Mechanism read from fumadocs-core dist buildFolder(): 'if (indexPath) { if (excludedPaths.has(indexPath)) delete node.index; else excludedPaths.add(indexPath); }'. CORRECTED A CLAIM IN THE INHERITED COMMIT MESSAGE: it said getBreadcrumbItems() 'drops the un-linkable ancestor'. It does not — it emits that ancestor with url:undefined; the drop happens in docsTrail() in apps/docs. Amended the message to say so. GATES DERIVED vs RUN: node scripts/pm/dispatch-gates.mjs --repo objectstack-ai/objectstack derived 56 entries. My worktree was 6 commits behind origin/main and the script warned STALE TREE (scripts/check-dts-emitted.mjs changed in that range), so I re-derived on a throwaway cmp worktree at current origin/main with the same one-line edit applied: 56 entries, comm both directions EMPTY — the family is identical, so the staleness is proven harmless rather than assumed so. Ran 39 gate commands + 4 CI self-test steps. comm -23 derived ran leaves 18: 16 are bracketed CI JOB STEPS, not local gate commands — 14 are workspace-wide build/typecheck/ESLint/install/store steps CI owns, and 2 (pr-automation.yml 'Label based on changed files', 'Re-read this PR's labels live') act on a live PR and are structurally un-runnable locally; the other 2 are spelling duplicates of gates I did run (node scripts/check-cross-package-test-inputs.mjs == pnpm check:cross-package-test-inputs; pnpm check:pm-half-states == node scripts/pm/check-half-states.mjs). THREE GATES ARE 'NOT MEASURED', NOT RED, each by its own printed declaration: check-test-completeness exit 3 'PREREQUISITE NOT MET — this gate grades a saved turbo run test log, and no log was named' (its own text says '⛔ It is not a red'); check-half-states exit 2 'sweep failed to run — GET .../comments -> HTTP 502' (no network egress from this runner); check:skill-examples exit 1 'packages/client-react/dist holds no .d.ts declarations — the package is not built'. For the third I attempted the build (pnpm --filter '@objectstack/client-react...' --filter '@objectstack/client...' build) and it failed after 6m45s in packages/runtime with 'error TS2307: Cannot find module @objectstack/rest' — an ordering artifact of MY narrowed filter (rest is an optional peer and so outside the closure I selected), not a defect and not reachable from a JSON nav file. CI builds the full workspace and measures all three. DECLARED NARROWING — ESLint: not run repo-wide. Three pieces of evidence, per the seat rule: (1) population read from eslint's own config resolution, not my guess — 'eslint --no-inline-config --format json content/docs/releases/meta.json' returns messages[0].message = 'File ignored because no matching configuration was supplied.', i.e. eslint declares this file outside its population; (2) file count read from --format json: 1 file, errorCount 0, fatalErrorCount 0; (3) config invariance — the diff is one JSON file and touches neither eslint.config.mjs nor anything it imports, and this file is not linted at all, so no untouched file's verdict can move. BYTES: node scripts/check-nul-bytes.mjs exit 0 ('scanned 7621 text file(s) ... no raw ASCII control bytes'), plus a manual grep -naP over the changed file for the wider control-byte class — no match.",
  "mcp_calls": "8 — create_pull_request 1, pull_request_read 2, issue_write 1 (labels), issue_read 2 (both REFUSED: 'API rate limit already exceeded'), search_issues 1 (REFUSED, same), add_issue_comment 1 (this comment). CHANNEL SWITCH DECLARED: the repo-scoped REST channel is unavailable from this seat — gh is ABSENT and a direct curl to api.github.com returned HTTP 502 'builtin injection failed (github)' — so every GitHub action here went over MCP. Within MCP the buckets differ: writes pass, while the GraphQL-backed reads (issue_read get_labels, search_issues) are rate-limited on the shared identity.",
  "open_questions": [],
  "out_of_scope_findings": [
    "COULD NOT FILE — handing to PM to file (the dedupe search and issue creation both need the GraphQL read bucket, which is rate-limited on the shared identity; per the seat rule a finding is never dropped because a channel broke, so it is reported here instead). FINDING: apps/docs/app/[lang]/docs/[[...slug]]/page.tsx carries a docblock, landed by PR #13710, that this PR makes factually stale. It reads: 'That was once 17 of the 35 meta.json files under content/docs, shortening 172 of 404 trails. 16 of the 17 were fixed producer-side (#12352) and 8 short trails remain, all under content/docs/releases/ — a directory AGENTS.md fences off, so its meta.json still lists \"index\". The condition is therefore live, just rare.' After this PR: 0 short trails remain and content/docs/releases/meta.json no longer lists \"index\", so both the count and the 'condition is live' conclusion are false. NOT fixed here on purpose — Zone 1 rule 3 forbids touching apps/docs/**, and rule 1 (one file) is the entire reason this PR is permitted inside the content/docs/releases/ fence; the bounded-in-place-fix exemption is overridden by that explicit adjudication. Suggested as an unassigned docs-comment issue against apps/docs, blocked-by / follow-on to this PR. Note the same docblock also still says the un-linkable ancestor is 'dropped' by getBreadcrumbItems(); measured, getBreadcrumbItems emits it with url:undefined and docsTrail() in that same file is what drops it — worth correcting in the same edit."
  ]
}

Generated by Claude Code

Closed by PR #13946, merged 2026-08-31T19:07:11Z. pm:blocking · pm:dispatched stripped by hand — 关闭即摘 pm:*.

⭐ One file, one line, no changeset — changed_files=1, deletions=1, commits=1 by GitHub's own count. That size was the legality: content/docs/releases/** is fenced unconditionally, and the only route in is the dedicated docs-only PR the fence itself names.

Measured, ⛔ not assumed: driving the real fumadocs-core loader over 405 pages, short breadcrumb trails go 8 → 0, and the landing page does not disappear — /docs/releases stays in getPages() and the folder gains indexUrl, flipping its sidebar header TRIGGER → LINK. ⭐ The before-leg was produced by re-inserting "index" in memory only, so there was no restore leg to get wrong.

⚠️ Now unblocked: #12352 carries Blocked-by: #13711 and this card is its blocker. ⛔ Left to the unlock scan rather than closed here.

⚠️ This PR made a docblock false and could not fix it — apps/docs/app/[lang]/docs/[[...slug]]/page.tsx:45-49 still says 8 short trails remain and the condition is live. Filed as #13949: the fence that made this fix possible is what stopped the fix carrying its own documentation correction.

Generated by Claude Code

docs(nav): stop listing "index" in content/docs/releases/meta.json (o…

Navigation Menu

Repository navigation

content/docs/releases/meta.json is the 17th file still listing "index" — the last 8 short breadcrumb trails, needing a dedicated docs-only PR #13711

What remains

The fix

Why the fence's reason does not reach this file — ⚠️ offered as analysis, ⛔ not as a licence

Verified, so the next seat need not re-derive

Why not just accept 8 permanently short trails

⛔ The fence, and why this card is the right way through it

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

content/docs/releases/meta.json is the 17th file still listing "index" — the last 8 short breadcrumb trails, needing a dedicated docs-only PR #13711

Description

What remains

The fix

Why the fence's reason does not reach this file — ⚠️ offered as analysis, ⛔ not as a licence

Verified, so the next seat need not re-derive

Why not just accept 8 permanently short trails

Activity

os-warren commented on Aug 31, 2026

⛔ The fence, and why this card is the right way through it

claude commented on Aug 31, 2026

claude commented on Aug 31, 2026

os-project-manager commented on Aug 31, 2026

claude commented on Aug 31, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions