Agent Onboarding Documentation
stableCategory: ai-signals · Methodology v4.5
Can an agent or developer find your integration docs starting from the homepage?
Signal Source
- Source
homepage HTML (developer/docs links + meta tags), well-known paths- Kind
- http_response
Score Bands
| Verdict | Condition |
|---|---|
| Pass | Comprehensive onboarding documentation — 3 or more developer/docs links (e.g. /developers, /documentation, /api-reference, /quickstart), OR both a getting-started/quickstart link and an API-reference/api-docs/openapi link |
| Partial | Basic onboarding documentation — 1 or 2 docs links, or only an ai-agent meta tag, with no comprehensive set |
| Fail | No discoverable onboarding docs links or agent meta tags |
Description
The Agent Onboarding Documentation parameter checks whether your homepage links to developer or agent integration docs. friendly4AI scans the homepage HTML for <a href> links to common documentation paths and for a <meta name="ai-agent"> tag, then scores the result 100 (pass), 50 (partial), or 0 (fail). Three or more docs links, or a getting-started guide paired with an API reference, earns a full pass.
What does this parameter measure?
Can an agent or developer find your integration docs starting from the homepage? That is the whole question. The scanner reads the homepage HTML and looks for <a href> links pointing at common documentation paths:
/developers,/developer,/developer-guide/documentation,/docs/api,/docs/getting-started/api-reference,/api-docs,/openapi,/swagger/getting-started,/quickstart
It also detects a <meta name="ai-agent"> tag. Editorial sections like /video/, /news/, or /health/ are excluded, so a newsroom path never gets mistaken for developer docs.
Why does it matter for AI-readiness?
A discoverable spec tells an agent what your API offers. Onboarding docs tell it how to start: how to authenticate, what the rate limits are, what a real request looks like. Link a getting-started guide and an API reference from the homepage, and both autonomous agents and the developers building integrations can serve themselves. Bury or skip this material, and every integration starts with guesswork. That slows adoption and breaks more calls than it should.
See also the related signals API Discoverability, Capability Metadata, and API Schema Quality for Agents.
How is it scored?
This is a gradient AI-Specific Signals parameter, and the scanner follows the published rubric:
- Pass (100) — Comprehensive documentation: three or more docs links, or a getting-started/quickstart link paired with an API-reference link.
- Partial (50) — Basic material only: one or two docs links, or just an
ai-agentmeta tag. - Fail (0) — No docs links or agent meta tags are discoverable.
The API-conditional checks in this category only fire once an API is detected. This one does not. It belongs to the AI-Specific Integration group and runs on every site, because onboarding docs help agents whether or not you publish a formal API spec.
How do you fix a low score?
- Publish a dedicated developer or agent documentation page, and link it from your homepage navigation or footer.
- Ship both a getting-started/quickstart guide and an API-reference page. That pairing alone reaches the pass tier.
- Document authentication steps, rate limits, and example requests so agents can serve themselves.
- Add a
<meta name="ai-agent" ...>tag that points to your docs as an extra machine-readable hint. - Keep developer docs out of editorial paths like
/news/or/video/, which the scanner excludes.
Version History
- Introduced
- v1
- Last changed
- v4.5
Key takeaways
- Signal: homepage HTML (developer/docs links + meta tags), well-known paths
- Category: AI-Specific Signals
- Passes when: Comprehensive onboarding documentation — 3 or more developer/docs links (e.g.…