# Vendor facets — GitBook. A hosted docs platform whose published sites now serve llms.txt,
# llms-full.txt, Markdown page twins and a per-site MCP server automatically, with OpenAPI blocks
# that carry a "Test it" console. The lift is thinner than the dedicated API-docs platforms: no
# api-catalog, no agent card, no customer skills, and a platform-graded docs MCP server.
vendor: gitbook
name: GitBook
website: https://gitbook.com
areas:
- documentation
registry_keys:
- gitbook
rubric_schema_version: 0.22.0
generated: '2026-09-25'
features_refreshed: '2026-09-25'
basis: measured
summary: >-
  GitBook earns a provider llms.txt and a docs MCP server without configuration, plus the ordinary documentation,
  reference, changelog and try-it pointers once the provider declares them. It stops there: GitBook serves
  no api-catalog, no A2A agent card and no customer-facing Agent Skills (its own SKILL.md teaches agents
  to edit GitBook, not to use the provider's API), and the per-site MCP server is one template, graded
  `platform`. Contract Quality is untouched because GitBook renders the OpenAPI it is given.
features:
- id: llm-ready-docs
  name: llms.txt, llms-full.txt and Markdown pages
  description: >-
    Every published site serves /llms.txt, /llms-full.txt and a .md version of each page, generated automatically.
  source: https://gitbook.com/docs/publishing-documentation/llm-ready-docs
  tier: all
- id: site-mcp
  name: MCP server for published docs
  description: >-
    Every published site exposes a GitBook-hosted MCP server at <site>/~gitbook/mcp for reading the docs,
    with OAuth discovery for authenticated sites.
  source: https://gitbook.com/docs/ai-for-your-readers/mcp-servers-for-published-docs
  tier: all
- id: mcp-server-card
  name: MCP server card
  description: GitBook's own docs site serves /.well-known/mcp/server-card.json (observed on gitbook.com/docs).
  source: https://gitbook.com/docs/.well-known/mcp/server-card.json
  tier: unknown
- id: openapi-blocks
  name: OpenAPI reference blocks with Test it
  description: >-
    Adds an OpenAPI spec to pages as interactive reference blocks with a "Test it" console and optional
    proxy for CORS-blocked APIs.
  source: https://gitbook.com/docs/create-content/openapi/guides/configuring-the-test-it-button
  tier: all
- id: custom-code-samples
  name: Custom code samples
  description: Displays custom code samples from the spec beside each endpoint.
  source: https://gitbook.com/docs/create-content/openapi/guides/adding-custom-code-samples
  tier: all
- id: updates-block
  name: Updates block (changelog) with RSS
  description: A dated, tagged updates block that turns any page into a changelog with an automatic RSS
    feed.
  source: https://gitbook.com/docs/create-content/blocks/updates
  tier: paid
- id: hosted-site
  name: Published docs site on a custom domain
  description: The hosted documentation site, on a gitbook.io subdomain, a custom domain or a subdirectory.
  source: https://gitbook.com/docs/publish/seo
  tier: all
- id: gitbook-skill
  name: GitBook's own SKILL.md
  description: An official SKILL.md that teaches coding agents to author GitBook content and drive the
    GitBook API.
  source: https://gitbook.com/docs/docs-as-code/ai-coding-assistants-and-skillmd
  tier: all
- id: adaptive-content
  name: Adaptive content and authenticated access
  description: Tailors pages to the reader from cookies, URL parameters, feature flags or an identity
    provider.
  source: https://gitbook.com/docs/publish/adaptive-content
  tier: unknown
- id: page-ask-hint
  name: Agent query hint in Markdown pages
  description: >-
    Markdown pages carry an instruction telling agents they can query the docs dynamically with an ask/goal
    request.
  source: https://gitbook.com/docs/publish/seo
  tier: unknown
maps:
- feature: llm-ready-docs
  check: llms_txt_published
  layer: composite
  provider_must: Publish the site on its own custom domain or subdirectory rather than a gitbook.io subdomain.
  note: >-
    Generated from the provider's own pages on the provider's domain, so it reads as the provider's llms.txt
    (rule 2's own-content, own-domain case).
  catalog_pass_rate: 0.351
  facet: discoverability
  points: 4
  baseline_pass_rate: 0.65
  cohort_pct: 76.6
  control_pct: 64.9
  delta_pp: 11.7
- feature: site-mcp
  check: mcp_server
  layer: agent_readiness
  grade: platform
  note: >-
    One GitBook-hosted template on every published site that reads docs and does not call the API; 0.22.0
    grades that `platform` (0.25 of 12). It is also disabled if page actions are off.
  points: 12
  baseline_pass_rate: 0.22
  cohort_pct: 26.4
  control_pct: 21.9
  delta_pp: 4.5
- feature: openapi-blocks
  check: console_or_sandbox
  layer: composite
  provider_must: Declare the reference page as a Console / Playground pointer.
  catalog_pass_rate: 0.089
  facet: developer_ergonomics
  points: 3
  baseline_pass_rate: 0.332
  cohort_pct: 31.9
  control_pct: 33.2
  delta_pp: -1.3
- feature: openapi-blocks
  check: api_reference_present
  layer: composite
  provider_must: Declare the reference page as an APIReference pointer.
  catalog_pass_rate: 0.222
  facet: developer_ergonomics
  points: 3
  baseline_pass_rate: 0.942
  saturated: true
  saturated_note: >-
    94% of providers with a contract, docs and a reference already earn this; the vendor cannot move it
    for most of its buyers.
  cohort_pct: 97.9
  control_pct: 94.1
  delta_pp: 3.8
- feature: hosted-site
  check: documentation_present
  layer: composite
  provider_must: Declare the site as a Documentation pointer.
  catalog_pass_rate: 0.453
  facet: developer_ergonomics
  points: 4
  baseline_pass_rate: 0.95
  saturated: true
  saturated_note: >-
    95% of providers with a contract, docs and a reference already earn this; the vendor cannot move it
    for most of its buyers.
  cohort_pct: 100.0
  control_pct: 94.9
  delta_pp: 5.1
- feature: updates-block
  check: change_log_present
  layer: composite
  provider_must: Use the updates block on a paid plan, keep posting, and declare the page as a ChangeLog
    pointer.
  catalog_pass_rate: 0.157
  facet: operational_transparency
  points: 6
  baseline_pass_rate: 0.438
  cohort_pct: 25.5
  control_pct: 44.0
  delta_pp: -18.5
- feature: custom-code-samples
  check: openapi_examples
  layer: agent_readiness
  grade: documented
  partial: true
  partial_note: >-
    The dimension reads examples inside the OpenAPI first; code samples GitBook renders reach only the
    Examples / Code Samples link fallback.
  points: 7
  baseline_pass_rate: 0.36
  cohort_pct: 24.5
  control_pct: 36.1
  delta_pp: -11.6
earns_nothing:
- feature: gitbook-skill
  check: agent_skills
  layer: agent_readiness
  why: >-
    The skill teaches agents to use GitBook itself; it says nothing about the provider's API and is not
    served as the provider's skill.
- feature: mcp-server-card
  check: well_known_published
  why: >-
    An MCP server card is not one of the four documents the check reads, and GitBook serves no api-catalog.
- feature: openapi-blocks
  check: examples_present
  why: Rendering the spec writes no examples back into it.
out_of_reach:
  checks:
  - agent_card
  - well_known_catalog
  - agent_skill_present
  - contract_present
  - info_complete
  - sdk_count_1
  - sdk_count_3
  - cli_present
  - postman_present
  - rules_present
  note: >-
    GitBook serves no agent card, api-catalog or customer skills, and does not author contracts, SDKs,
    a CLI, collections or rulesets.
unscored_practice:
- feature: llm-ready-docs
  why: >-
    Markdown page twins and llms-full.txt are agent-facing formats no check separates from a stub llms.txt.
- feature: page-ask-hint
  why: An in-content instruction telling agents how to query the docs; no check reads agent directives.
- feature: adaptive-content
  why: >-
    Reader-specific docs (personalized keys and content) are real onboarding practice with no scored signal.
surface:
  discoverability:
    reachable: 4.0
    total: 54
  operational_transparency:
    reachable: 6.0
    total: 38
  developer_ergonomics:
    reachable: 10.0
    total: 42
  agent_readiness:
    reachable: 4.8
    total: 139
hard_rule: >-
  A model, not a score. Adopting this vendor changes a provider's Kin Score only when the provider publishes
  the resulting artifacts on its own surface; nothing here writes a score, and no sponsorship or partnership
  can.
method: searched
source:
- https://gitbook.com/docs/.well-known/mcp/server-card.json
- https://gitbook.com/docs/ai-for-your-readers/mcp-servers-for-published-docs
- https://gitbook.com/docs/create-content/blocks/updates
- https://gitbook.com/docs/create-content/openapi/guides/adding-custom-code-samples
- https://gitbook.com/docs/create-content/openapi/guides/configuring-the-test-it-button
- https://gitbook.com/docs/docs-as-code/ai-coding-assistants-and-skillmd
- https://gitbook.com/docs/publish/adaptive-content
- https://gitbook.com/docs/publish/seo
- https://gitbook.com/docs/publishing-documentation/llm-ready-docs
measured:
  cohort:
    method: vendors-catalog.json detections (CNAME / header / URL shape / markup), never a name match
    detected: 112
    in_baseline: 47
  control:
    basis: providers earning contract_present + documentation_present + api_reference_present, minus the
      cohort
    n: 5169
  metric: >-
    cohort_pct / control_pct = mean share of the check's points earned (derived and platform credit weighted),
    x100
  measured_on: '2026-09-25'
  composite_mean:
    cohort: 45.5
    control: 48.1
  agent_readiness_mean:
    cohort: 31.2
    control: 32.5
  status: measured
  caveat: >-
    A cohort delta is association, not cause: customers choose a vendor for reasons that also move their
    score. Read it beside the capability map, never instead of it.
simulation:
  simulated_on: '2026-09-25'
  rubric: 0.23.0
  population: providers publishing a contract (contract_present earned), replayable exactly
  providers: 8977
  providers_unreplayable: 987
  providers_moved: 8756
  conditional_rows: excluded (they depend on what the API already does)
  composite_lift:
    median: 3.6
    p75: 5.6
    p90: 5.7
    max: 7.6
    mean_among_movers: 3.9
  agent_readiness_lift:
    median: 2.2
    p75: 3.4
    p90: 3.5
    max: 4.0
    mean_among_movers: 2.7
  facet_lift_median_among_movers:
    discoverability: 7.4
    operational_transparency: 15.8
    developer_ergonomics: 11.9
  composite_band_moves:
    thin -> developing: 1464
    developing -> strong: 551
    emerging -> thin: 326
    strong -> exemplar: 118
    minimal -> emerging: 3
  agent_readiness_band_moves:
    agent-aware -> agent-ready: 1006
    agent-ready -> agent-native: 87
  method: >-
    each provider's own kin/checks file, the vendor's maps at their stated credit, the scorer's composite
    formula; from -> to, nothing written