# API reference

The AutoSugges runtime HTTP API — GET /v1/query, GET /v1/bootstrap, POST /v1/selection, and the complete typed error vocabulary, with real curl examples.

- [GET /v1/query — the autocomplete endpoint](https://docs.autosugges.com/api-reference/query.md): Call the runtime directly for pre-ranked autocomplete suggestions — the endpoint every SDK binding calls underneath.
- [GET /v1/bootstrap — session policy and version resolution](https://docs.autosugges.com/api-reference/bootstrap.md): Resolve a publishable key to its list identity, active version hash and effective query policy in one call, before the first query.
- [POST /v1/selection — report which suggestion was chosen](https://docs.autosugges.com/api-reference/selection.md): Tell AutoSugges which suggestion a user picked, so future ranking for that list can improve. Entirely optional and fire-and-forget.
- [Mint an access token from your server](https://docs.autosugges.com/api-reference/token.md): Exchange a secret key for a one-hour access token on your backend, then query a private list from the browser with it.
- [Error codes](https://docs.autosugges.com/api-reference/errors.md): Every typed error code the AutoSugges runtime API returns — HTTP status, retry posture, the exact message you receive, and what to do about each.
- [Call the control-plane API with a secret key](https://docs.autosugges.com/api-reference/control-plane.md): Read and change your workspace’s lists from a server with a secret key: cursor pagination, idempotency keys and If-Match concurrency.

Machine-readable description: https://docs.autosugges.com/openapi.json (OpenAPI 3.1)

---
Canonical page: https://docs.autosugges.com/api-reference