feat(seo): noindex collections and repo hubs, drop their sitemaps

app/pages/@[login]/[slug].vue

@@ -1,6 +1,6 @@
 <script setup lang="ts">
+import { entityRobots } from '#shared/entity-robots'
 import { githubAvatarProxyUrl } from '#shared/image-proxy'
-import { isNoindexCollection } from '#shared/retired-collections'
 
 const route = useRoute()
 const login = computed(() => String(route.params.login))
@@ -35,8 +35,8 @@ useSeoMeta({
   // "Apple Apps · @harlan-zw · skilld · skilld".
   title: () => collection.value ? `${collection.value.name} · @${login.value}` : 'Collection',
   description: () => collection.value?.preamble ?? `Skill collection by @${login.value}`,
-  // The featured band and curator signals need these rows live; search does not need the pages.
-  robots: () => isNoindexCollection(login.value, slug.value) ? 'noindex,follow' : 'index,follow',
+  // Owner decision 2026-10-01: collections stay live and linked, never indexed.
+  robots: entityRobots({ _tag: 'collection' }) ?? undefined,
 })
 
 const canonicalUrl = computed(() => `https://skilld.dev/@${login.value}/${slug.value}`)

app/pages/@[login]/index.vue

@@ -1,5 +1,6 @@
 <script setup lang="ts">
 import type { OrgProfile } from '#layers/registry/server/api/orgs/[owner].get'
+import { entityRobots } from '#shared/entity-robots'
 import { avatarProxyUrl } from '#shared/image-proxy'
 import { authorBadgeInput } from '../../utils/author-badge'
 
@@ -201,6 +202,8 @@ useSeoMeta({
     : `${displayName.value} (@${login.value})`,
   description: () => `${skills.value.length} ${skills.value.length === 1 ? 'skill' : 'skills'} and ${collections.value.length} ${collections.value.length === 1 ? 'collection' : 'collections'} from @${login.value} on skilld.`,
   ogUrl: canonicalUrl,
+  // Owner decision 2026-10-01: profiles stay live and linked, never indexed.
+  robots: entityRobots({ _tag: 'author-profile' }) ?? undefined,
 })
 
 useHead({

docs/work/EXECUTE-seo-recovery.md

@@ -160,6 +160,7 @@ Check each PR with `gh pr view <n> --repo skilld-dev/skilld.dev`.
 | 5 | Crawl waste: `www` 301; 75 to 19 JS preloads; HTML edge cache | [#316](https://github.com/skilld-dev/skilld.dev/pull/316), [#323](https://github.com/skilld-dev/skilld.dev/pull/323), [#328](https://github.com/skilld-dev/skilld.dev/pull/328), [#329](https://github.com/skilld-dev/skilld.dev/pull/329) |
 | 6 | Real 404 and 410 for missing Skills; 503 with `Retry-After` when the lookup fails | [#317](https://github.com/skilld-dev/skilld.dev/pull/317) |
 | 7 | 40 URL panel | [seo-recovery-panel.json](seo-recovery-panel.json) |
+| 8 | Author profiles, collections, owner hubs and multi-Skill repository hubs render `noindex,follow`. The `authors` and `sources` sitemaps are gone (owner decision, 2026-10-01). A single-Skill repository hub is the Skill's page and stays in the skills sitemap when the trending admission rule admits it | this change |
 
 Checks:
 

layers/registry/app/pages/gh/[owner]/[repo]/index.vue

@@ -3,6 +3,7 @@ import type { OrgProfile } from '../../../../../server/api/orgs/[owner].get'
 import type { RepoSourceProfile } from '../../../../../server/api/repos/[owner]/[repo].get'
 import type { RepoHistoryResponse } from '../../../../../server/api/repos/[owner]/[repo]/history.get'
 import type { RepoRouteResolution } from '../../../../../server/api/repos/[owner]/[repo]/route-target.get'
+import { entityRobots } from '#shared/entity-robots'
 import { avatarProxyUrl } from '#shared/image-proxy'
 import { resolveMissingRepoRedirect } from '../../../../utils/missing-repo-recovery'
 import { parseRepoSkillSort, REPO_SKILL_SORT_OPTIONS, sortRepoSkills } from '../../../../utils/repo-skill-layout'
@@ -299,10 +300,8 @@ const skillDescription = computed(() => {
 useSeoMeta({
   title: () => skillTitle.value,
   description: () => skillDescription.value,
-  // Indexable only when the hub renders its source. A renamed repository comes
-  // back under a new identity, `repoSource` is null, and the page shows
-  // "Source not found", so it is noindex. The API applies the same rule.
-  robots: () => repoSource.value?.seoIndexable ? 'index,follow' : 'noindex,follow',
+  // One decision: a hub rendering SkillDetail leaves robots to the Skill page.
+  robots: () => entityRobots({ _tag: 'repo-hub', renderedAsSkill: flatSkillName.value !== null }) ?? undefined,
   ogTitle: () => skillTitle.value,
   ogDescription: () => skillDescription.value,
   twitterTitle: () => skillTitle.value,

layers/registry/app/pages/gh/[owner]/index.vue

@@ -1,5 +1,6 @@
 <script setup lang="ts">
 import type { OrgProfile } from '../../../../server/api/orgs/[owner].get'
+import { entityRobots } from '#shared/entity-robots'
 import { avatarProxyUrl } from '#shared/image-proxy'
 import { resolveOwnerProfileHandoff } from '../../../utils/owner-profile-handoff'
 
@@ -131,7 +132,8 @@ useSeoMeta({
     return `${data.value.totalSkills} agent skills ${verb} ${data.value.displayName}.`
   },
   ogUrl: canonicalUrl,
-  robots: () => data.value?.seoIndexable ? 'index,follow' : 'noindex,follow',
+  // Owner hubs stay live and linked, never indexed (owner decision 2026-10-01).
+  robots: entityRobots({ _tag: 'owner-hub' }) ?? undefined,
   twitterCard: 'summary_large_image',
 })
 

layers/registry/server/api/orgs/[owner].get.ts

@@ -8,7 +8,6 @@ import { TAG_BY_SLUG } from '../../jobs/taxonomy'
 import { resolveRepoSourceIdentitiesForOwner } from '../../utils/repo-source-identity'
 import { getGeneratedBatch } from '../../utils/skill-generated'
 import { querySkills } from '../../utils/skills-registry'
-import { isTrustedAuthorOwner } from '../../utils/trusted-author-sources'
 
 export type OrgKind = 'org' | 'user'
 
@@ -42,7 +41,6 @@ export interface OrgProfile {
   skills: RegistrySkill[]
   lastSyncedAt: number | null
   syncStatus: 'ok' | 'failed' | 'never' | null
-  seoIndexable: boolean
   fetchedAt: string
 }
 
@@ -279,7 +277,6 @@ async function loadOrgProfile(event: H3Event, owner: string): Promise<OrgProfile
     skills: registryResult.items,
     lastSyncedAt,
     syncStatus,
-    seoIndexable: isTrustedAuthorOwner(owner),
     fetchedAt: new Date().toISOString(),
   }
 

layers/registry/server/api/repos/[owner]/[repo].get.ts

@@ -1,10 +1,8 @@
 import { getDB } from '#server/utils/db'
-import { hubRendersSource } from '#shared/repo-identity'
 import { isRegistrySkillPath } from '#shared/skill-path'
 import { getRepo, getTree, resolveGithubBindings } from '../../../utils/github-client'
 import { resolveRepoSourceIdentity } from '../../../utils/repo-source-identity'
 import { buildUnavailableRepoSourceProfile } from '../../../utils/repo-source-profile'
-import { isTrustedAuthorRepo } from '../../../utils/trusted-author-sources'
 
 export interface RepoSourceProfile {
   owner: string
@@ -21,7 +19,6 @@ export interface RepoSourceProfile {
   skillFileScanStatus: 'ok' | 'unavailable' | 'truncated'
   skillFileCount: number
   skillFiles: string[]
-  seoIndexable: boolean
 }
 
 export default defineCachedEventHandler(async (event) => {
@@ -32,7 +29,6 @@ export default defineCachedEventHandler(async (event) => {
 
   const owner = ownerParam.toLowerCase()
   const repo = repoParam.toLowerCase()
-  const trusted = isTrustedAuthorRepo(owner, repo)
   const db = getDB(event)
   const source = await resolveRepoSourceIdentity(db, { owner, repo })
   const bindings = resolveGithubBindings(event.context.platform.env)
@@ -46,18 +42,12 @@ export default defineCachedEventHandler(async (event) => {
       'outcome': 'unavailable',
       'upstream.status': repoRes.status || null,
     }))
-    return {
-      ...buildUnavailableRepoSourceProfile(owner, repo),
-      seoIndexable: trusted,
-    } satisfies RepoSourceProfile
+    return buildUnavailableRepoSourceProfile(owner, repo) satisfies RepoSourceProfile
   }
 
   const meta = repoRes.data
   const repoOwner = meta.owner.login
   const repoName = meta.name
-  // A renamed repository comes back under its new identity, and the hub page
-  // then renders "Source not found". Only a hub that renders its source is indexable.
-  const seoIndexable = trusted && hubRendersSource({ owner, repo }, { owner: repoOwner, repo: repoName })
   const treeRes = await getTree(repoOwner, repoName, meta.default_branch, bindings)
   const skillFiles = (treeRes.data?.tree ?? [])
     .filter(entry => entry.type === 'blob' && isRegistrySkillPath(entry.path))
@@ -84,7 +74,6 @@ export default defineCachedEventHandler(async (event) => {
     skillFileScanStatus,
     skillFileCount: skillFiles.length,
     skillFiles,
-    seoIndexable,
   } satisfies RepoSourceProfile
 }, {
   maxAge: 60 * 15,

nuxt.config.ts

@@ -721,8 +721,7 @@ export default defineNuxtConfig({
   sitemap: {
     // `/learn` index is a 55-word card list, noindex since 2026-08-22
     // (GOOGLE_RECOVERY.md). Articles stay indexable and sitemap-listed; only
-    // the bare index leaves. Global so no child sitemap can re-adopt it
-    // (authors.xml was listing it via an app-source merge quirk).
+    // the bare index leaves. Global so no child sitemap can re-adopt it.
     exclude: ['/learn'],
     sitemaps: {
       pages: {
@@ -743,14 +742,10 @@ export default defineNuxtConfig({
         includeAppSources: false,
         chunks: 10000,
       },
-      authors: {
-        sources: ['/api/__sitemap__/authors'],
-        includeAppSources: false,
-      },
-      sources: {
-        sources: ['/api/__sitemap__/trusted-authors'],
-        includeAppSources: false,
-      },
+      // `authors` and `sources` removed 2026-10-01 (owner decision): author
+      // profiles, collections, owner hubs and multi-Skill repository hubs render
+      // `noindex,follow`. A single-Skill repository hub is the Skill's own page,
+      // so the skills sitemap lists it when the trending admission rule admits it.
       // `orgs` removed: it listed every owner hub (/gh/<owner>) unconditionally,
       // but those pages render noindex,follow. Advertising noindex URLs in the
       // sitemap was the bulk of GSC "Crawled – currently not indexed" (~8k) and

shared/entity-robots.ts

@@ -0,0 +1,23 @@
+/**
+ * The robots directive for pages that are not Skill pages.
+ *
+ * Owner decision, 2026-10-01: author profiles, collections, owner hubs and
+ * multi-Skill repository hubs stay live and linked, and are never indexed. No
+ * sitemap lists them, so the sitemap equals the indexable set.
+ *
+ * A single-Skill repository hub is the Skill's own page. It returns `null`
+ * here, so the page emits no robots tag of its own and the Skill page decision
+ * (`resolveSkillPageState`) is the only one. Two competing tags would let a
+ * default noindex override an admitted Skill.
+ */
+export type EntityPage
+  = | { _tag: 'author-profile' }
+    | { _tag: 'collection' }
+    | { _tag: 'owner-hub' }
+    | { _tag: 'repo-hub', renderedAsSkill: boolean }
+
+export function entityRobots(page: EntityPage): 'noindex,follow' | null {
+  if (page._tag === 'repo-hub' && page.renderedAsSkill)
+    return null
+  return 'noindex,follow'
+}

shared/retired-collections.ts

@@ -8,9 +8,10 @@
  * D1 on purpose. `/api/collections/featured` reads only featured rows, so
  * retiring them would empty the homepage band. Retiring them would also drop
  * `curator_count` and `curator_reason_count`, which count live collections only
- * and gate indexability for some Skills. Their pages render `noindex,follow` and
- * stay out of the authors sitemap. Revisit once other collections are featured
- * and the curator signals no longer depend on these rows.
+ * and gate indexability for some Skills. Since 2026-10-01 every collection page
+ * renders `noindex,follow`, so the list only records which rows must stay live.
+ * Revisit once other collections are featured and the curator signals no
+ * longer depend on these rows.
  */
 export interface CollectionRef {
   author: string
@@ -57,12 +58,3 @@ function has(list: readonly CollectionRef[], author: string, slug: string): bool
 export function isRetiredCollection(author: string, slug: string): boolean {
   return has(RETIRED_COLLECTIONS, author, slug)
 }
-
-export function isNoindexCollection(author: string, slug: string): boolean {
-  return has(NOINDEX_COLLECTIONS, author, slug)
-}
-
-/** Drops collections that must stay out of the sitemap. Runs before the migration lands too. */
-export function withoutUnindexedCollections<T extends { author_login: string, slug: string }>(rows: readonly T[]): T[] {
-  return rows.filter(row => !isRetiredCollection(row.author_login, row.slug) && !isNoindexCollection(row.author_login, row.slug))
-}