docs(nav): stop listing "index" in meta.json pages — it detaches the folder index (#12352)

Fumadocs attaches a folder's index.mdx as that folder's tree `index` node only
when the folder's meta.json does NOT list "index" in `pages`. Listing it makes
the page an ordinary child and leaves the folder node with a name and no url,
which degrades both surfaces built from that node: `getBreadcrumbItems()` drops
the un-linkable ancestor, and the sidebar renders an inert `SidebarFolderTrigger`
instead of a `SidebarFolderLink`.

Removes "index" from 16 of the 17 meta.json files that listed it.
content/docs/releases/ is fenced by AGENTS.md and is left untouched.

Measured on the real fumadocs-core 16.14.4 loader over content/docs:
short breadcrumb trails 172 -> 8 (the 8 are all under /docs/releases), and the
sidebar delta is exactly 16 folder headers TRIGGER -> LINK plus 16 index
children leaving the child list, with no other entry moved.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Pk26oZ12t5N1hwGW1m1MgC

Author: claude

.changeset/docs-meta-index-detaches-folder-index.md

@@ -0,0 +1,41 @@
+---
+"@objectstack/docs": patch
+---
+
+fix(docs): stop listing `"index"` in `meta.json` `pages` — it detaches the folder index and shortens 164 breadcrumb trails (#12352)
+
+Fumadocs attaches a folder's `index.mdx` as that folder's tree `index` node
+**only when the folder's `meta.json` does not list it** in `pages`. Listing it
+makes the page an ordinary child instead, and the folder node reaches every tree
+consumer with a `name` and no `url` — `loader-*.js`, `buildFolder()`:
+
+```js
+if (indexPath) {
+  if (excludedPaths.has(indexPath)) delete node.index;   // "index" was listed
+  else excludedPaths.add(indexPath);
+}
+```
+
+Two surfaces read that one node, and both were degraded:
+
+- **Breadcrumb.** `getBreadcrumbItems()` links a folder crumb to `item.index?.url`,
+  so an un-linkable ancestor is dropped rather than emitted name-only (Google
+  requires `item` on every `BreadcrumbList` entry but the last). 172 of 404 doc
+  pages advertised a two-level site structure they do not have.
+- **Sidebar.** `node.index ? SidebarFolderLink : SidebarFolderTrigger` — the
+  section header was inert text, and the section's own overview page sat below it
+  as a child, in six cases under a label identical to the header's.
+
+`"index"` is removed from 16 of the 17 `meta.json` files that listed it. It was
+the first `pages` entry in 15 of them and the first entry after the
+`---Start Here---` separator in `getting-started`, so no other entry's position
+depends on it: the measured tree delta is exactly 16 folder headers going
+`TRIGGER` → `LINK` and 16 index children leaving the child list, with every
+removed child's URL now the header's `href` and no other line moved.
+
+Short trails: **172 → 8**. The remaining 8 are `content/docs/releases/`, which
+this PR does not touch — that directory is fenced by AGENTS.md, and its
+`meta.json` still lists `"index"`.
+
+No consumer-side change: `app/[lang]/docs/[[...slug]]/page.tsx` reconstructs no
+URLs, deliberately, so a producer defect of this shape stays visible.

apps/docs/app/[lang]/docs/[[...slug]]/page.tsx

@@ -39,23 +39,21 @@ type DocPage = NonNullable<ReturnType<typeof source.getPage>>;
  * `index` node. ⛔ Nothing here splits `page.url` on `/`, and nothing constructs a
  * URL the loader has not already produced.
  *
- * ⚠️ **An ancestor arrives without a URL more often than not, and this drops it.**
- * Measured against a local production build, not inferred: fumadocs attaches a
- * folder's `index.mdx` as that folder's `index` node only when the folder's
- * `meta.json` does **not** list `"index"` in `pages`. 17 of the 35 `meta.json`
- * files under `content/docs` do list it, so their folder nodes carry a
- * `name` and no `url`, and 172 of 403 doc pages therefore ship a trail that skips
- * its section. Confirmed causally by deleting that one line from
- * `content/docs/data-modeling/meta.json` and rebuilding: the section crumb
- * appeared, linked, while an untouched control section stayed short.
+ * ⚠️ **An ancestor can arrive without a URL, and this drops it.** Fumadocs
+ * attaches a folder's `index.mdx` as that folder's `index` node only when the
+ * folder's `meta.json` does **not** list `"index"` in `pages`; a folder that
+ * lists it reaches this walk with a `name` and no `url`. 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.
  *
- * ⛔ The missing URL is deliberately **not** reconstructed here. The folder's index
- * page exists, is in the sitemap and answers 200 — the defect is in the content
- * config that hides it from the tree, not in this consumer, and a lookup that
- * re-derived it would make a producer bug invisible and permanent. Google requires
- * `item` on every crumb but the last, so a name-only crumb is not an option
- * either. Filed separately; when it lands, these trails complete with no change to
- * this file.
+ * ⛔ The missing URL is deliberately **not** reconstructed here — that fence is
+ * the reason #12352 was fixable at all. The folder's index page exists, is in the
+ * sitemap and answers 200 — the defect is in the content config that hides it
+ * from the tree, not in this consumer, and a lookup that re-derived it would make
+ * a producer bug invisible and permanent. Google requires `item` on every crumb
+ * but the last, so a name-only crumb is not an option either.
  *
  * Two things the tree cannot supply are added around it, both from data this
  * page already holds:

content/docs/ai/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "AI",
   "pages": [
-    "index",
     "connect-mcp",
     "agents",
     "actions-as-tools",

content/docs/api/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "API & SDK",
   "pages": [
-    "index",
     "data-api",
     "metadata-api",
     "plugin-endpoints",

content/docs/automation/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "Automation",
   "pages": [
-    "index",
     "hooks",
     "hook-bodies",
     "flows",

content/docs/capabilities/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "What Can It Do?",
   "pages": [
-    "index",
     "data",
     "views",
     "forms",

content/docs/concepts/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "Core Concepts",
   "pages": [
-    "index",
     "architecture",
     "metadata-driven",
     "metadata-lifecycle",

content/docs/data-modeling/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "Data Modeling",
   "pages": [
-    "index",
     "schema-design",
     "objects",
     "fields",

content/docs/deployment/meta.json

@@ -1,7 +1,6 @@
 {
   "title": "Deployment & Operations",
   "pages": [
-    "index",
     "self-hosting",
     "production-readiness",
     "backup-restore",

content/docs/getting-started/meta.json

@@ -2,7 +2,6 @@
   "title": "Get Started",
   "pages": [
     "---Start Here---",
-    "index",
     "how-ai-development-works",
     "build-with-claude-code",
     "your-first-project",