# Vertical manifest v1 — `kind` selects the runtime in src/extract/vertical/.
version: 1
order: 98
name: docsite
# html-extract: fetch matched URL as HTML → fields (selectors, rules).
kind: html-extract
description: Documentation site structured extraction (API ref, changelog, FAQ, compatibility).
urlPatterns:
  - https://:host/docs/:path*
  - https://:host/api/:path*
  - https://*.readthedocs.io/:path*
  - https://*.gitbook.io/:path*
  - https://*.gitbook.com/:path*
  - https://gitbook.com/:path*
  - https://developer.mozilla.org/:locale/docs/:path*
  - https://:host/:path*
# Runtime requirements for the scrape host.
requirements:
  requiresBrowser: true
  requiresLLM: false
  requiresCloud: false
# Declared output facets (discovery / tooling).
capabilities:
  - api_reference
  - changelog
  - faq
  - compatibility_table
source: builtin
# HTML field rules (selectors, meta, headingSections, platform-specific kinds).
fields:
  platform:
    kind: docsitePlatform
  version:
    kind: docsiteVersion
  breadcrumbs:
    kind: breadcrumbs
    selectors:
      - 'nav[aria-label*=breadcrumb i] a, nav[aria-label*=breadcrumb i] li, .breadcrumbs a, .breadcrumbs li, .breadcrumbs span'
      - '.wy-breadcrumbs a, .wy-breadcrumbs li, .breadcrumb a, .breadcrumb li'
  title:
    selectorText:
      - main h1
      - article h1
      - h1
      - title
  summary:
    meta:
      - description
      - 'og:description'
  sections:
    kind: headingSections
    contentChars: 1200
    rootSelectors:
      - article
      - main
      - .theme-doc-markdown
      - '.rst-content .document'
      - .markdown-section
  apiSignature:
    kind: mdnSignature
  source:
    object:
      provider:
        value: docsite
      finalUrl:
        value: "{{url}}"