Publishing a scraper as an API endpoint

1. 1Build and test a scraper until its output looks right.
2. 2Open Endpoints and create one: pick a slug, a method (GET or POST), the scraper behind it, and optional parameter and output schemas.
3. 3Choose freshness, serve from cache or always scrape fresh, and a cache TTL up to 7 days.
4. 4Choose sync (wait for the result) or async (get a run id back and poll for it).
5. 5Create an API key with the endpoints:invoke scope and call it.

Calling your endpoint

curl https://app.manypi.com/v1/e/your-slug \
  -H "Authorization: Bearer mpi_your_api_key"

Polling an async invocation

curl https://app.manypi.com/v1/e/your-slug/results/RUN_ID \
  -H "Authorization: Bearer mpi_your_api_key"

OpenAPI spec

Every workspace gets a generated OpenAPI 3.1 document covering all of its endpoints, point your codegen or your agent framework straight at it.

curl https://app.manypi.com/v1/openapi.json \
  -H "Authorization: Bearer mpi_your_api_key"

Slugs are scoped to your workspace, so /v1/e/pricing is yours alone. Most cached and quick scrapes answer synchronously; slower ones return a run id to poll.