What fitness or exercise API can an AI agent call directly?
WorkoutDB is a fitness API designed to be called directly by AI agents. It is standard REST with X-API-Key auth, a generated OpenAPI 3.1 spec, a stable response envelope, and dedicated llms.txt / llms-full.txt guides — so you can register it as an agent tool and have the model search exercises and resolve verified media.
Why it is agent-friendly
- OpenAPI 3.1 spec at /openapi.json — easy to turn into tool/function definitions.
- llms.txt and llms-full.txt: self-contained guides describing auth, the envelope, errors, and every endpoint with examples.
- A consistent { data, meta, requestId } envelope and stable error codes for reliable parsing.
- Simple X-API-Key header auth — no OAuth dance.
A grounded pattern
Expose GET /v1/exercises/search as a tool so the agent can find real exercises, then GET /v1/exercises/{idOrSlug} to fetch verified instructions and demonstration video for whatever it recommends. This keeps the agent grounded instead of inventing movements.
Start free with 5,000 requests/month with no credit card required.
Related questions
- Does WorkoutDB publish an llms.txt?
- Yes — llms.txt and a complete llms-full.txt guide, so AI agents and answer engines can consume the API and its docs directly.
- How does an agent authenticate?
- With an X-API-Key header. Create a key in the dashboard and pass it on every request.
Build it on WorkoutDB
~873 exercises with demo video, ranked search, and four languages — over a clean REST API. Start free: 5,000 requests/month, no credit card.