AgentFit API documentation
Audit any public API documentation site against a 30-criteria AI-readiness rubric over HTTP, CLI, or MCP.
What AgentFit is
AgentFit audits public API documentation for AI-readiness — how well an LLM or autonomous agent can read your docs and call your API correctly without guessing. Give it a base URL and it returns a deterministic JSON report scoring the site from 0 to 100 across six categories: Discovery, Page artifacts, API spec, Content, Hygiene, and Agent Surface. Every criterion is backed by concrete HTTP evidence, and there are no LLM calls at run time — the score is reproducible byte-for-byte.
You can run an audit three ways: the HTTP API documented here, the
agentfit command-line tool, or the remote MCP server that connects AgentFit
directly into Claude, Cursor, and other agent clients. This site documents all three.
Audit a docs site now
Paste a documentation base URL — the form runs a live audit on agentfit.dev.
Quickstart
Start an asynchronous audit and poll for the scored report:
curl -s -X POST 'https://agentfit.dev/audit?async=true' \
-H 'Content-Type: application/json' \
-d '{"base_url":"https://docs.stripe.com"}'
# {"run_id":"…","status":"pending","poll_url":"/api/public/audit/…","share_url":"/r/…"}
curl -s 'https://agentfit.dev/api/public/audit/RUN_ID' | jq '.status, .total_score'
The machine-readable contract is published as an OpenAPI 3.1 document at
/openapi.yaml (and as JSON at
/openapi.json). The full reference, authentication,
error model, and language examples are linked from
llms-full.txt.
Always poll the durable /api/public/audit/{id}
endpoint. The legacy /audit/{id} in-memory poll path is deprecated
and not part of the versioned contract.
Connect from an agent (MCP)
AgentFit hosts a remote, OAuth-protected MCP server. Add it to Claude Code with no sign-up — the OAuth step auto-approves:
claude mcp add --transport http agentfit https://docs.agentfit.dev/mcp
Tools: audit(base_url, format) and get_report(run_id), scope
agentfit:audit. The discovery surface lives at
/.well-known/oauth-protected-resource.
Documentation
- Quickstart — run your first audit in under a minute
- Start an audit —
POST /audit - Get an audit — poll a run and read the report
- List audit history · Diff two audits · Browse the corpus
- Authentication and rate limits · Errors · Glossary
- MCP server · Changelog · Terms & license · OpenAPI