# AgentFit

> AI-readiness audit for API documentation. Deterministic 0-100 score
> across 26 criteria, five categories (discovery, machine-readable,
> structure, depth, hygiene). No LLM at runtime — every score traces
> to a concrete HTTP fetch or parsed structure.

## How it works

- [Audit homepage](https://agentfit.dev/): submit a docs URL, get a score in ≤30s
- [Public report archive](https://agentfit.dev/browse): recently audited sites
- [Top scores](https://agentfit.dev/browse/top): leaderboard across the corpus
- [Stats](https://agentfit.dev/browse/stats): corpus size, score distribution
- [Search](https://agentfit.dev/browse/search): find audits by host

## API

- [OpenAPI 3.1 spec](https://agentfit.dev/openapi.yaml): full machine-readable contract
- POST `/audit` body `{"base_url":"..."}` — synchronous audit, returns full JSON report
- POST `/audit?async=true` — returns 202 + `{job_id, poll_url, share_url}`
- GET `/audit/{id}` — poll async result; 200 when done, 202 while running
- GET `/healthz` — liveness + feature flags (ml, headless availability)

## Operations

- [Crawler info](https://agentfit.dev/bot): User-Agent strings, rate limits, opt-out
- [Privacy](https://agentfit.dev/privacy): personal-data stance, audit-removal request
- [Terms of use](https://agentfit.dev/terms): public-report consent, no-warranty