answerline-dev Async API

openapi: 3.2.0
info:
  title: Answerline Async API
  description: 'Answerline tracks how AI assistants and search engines answer, across engines and regions. Each request comes back as structured JSON holding the engine''s answer: the parsed text, its cited sources and whatever else the engine returned with them, geo-targeted to a country you choose (which countries are available depends on the engine; see coverage).


    ## Authentication


    Pass your API key in the header `Authorization: Bearer `. You can create keys yourself at https://dashboard.answerline.dev/keys, and each new account starts with free requests, no card needed. A single key works for all endpoints; keys cannot currently be limited to particular scopes.


    ## Versioning and deprecation


    The URL path carries the version. All endpoints sit under `/v1/`; no header or query parameter selects a version.


    Changes inside a version only add things: new endpoints, optional request fields and response fields may be introduced at any moment. Clients have to skip response fields they don''t know. Anything breaking ships under a new path (`/v2/`) and never replaces an existing version.


    Deprecation is announced in the endpoint''s responses themselves, not just in a changelog: the `Deprecation` header (RFC 9745) gives the date it became deprecated, `Sunset` (RFC 8594) gives the date it will stop answering, and the replacement is given by a `Link` header carrying `rel="deprecation"`. For generally available endpoints, those two dates are at least six months apart. No endpoint is currently deprecated.


    ## Rate limits


    There is no per-account request rate limit and no cap on how many requests run at once. Each response still carries `X-RateLimit-Limit` and `X-RateLimit-Remaining` for clients that read them: both are deprecated and always 100000, meaning no limit. The concurrency and USD balance headers report your account''s state. Your async queue is bounded only by the memory its tasks take: a submission that does not fit is refused whole with `QUEUE_LIMIT_EXCEEDED`, whose details say how many more tasks of that size fit. A synchronous call that cannot start now is refused with `CONCURRENT_LIMIT_EXCEEDED`, which goes away once your running jobs complete. `Retry-After` is set only on `SERVICE_OVERLOADED` and on a `RATE_LIMIT_EXCEEDED` sent because one address presented too many unknown keys. No IETF `RateLimit-*` headers are sent.


    ## Errors


    All 4xx and 5xx responses share the `Error` envelope: `{ "error": { "code", "message", "details", "timestamp" } }`, except validation errors, which use `{ "success": false, "error", "details": [{ "field", "message" }] }`. Clients should branch on `code`, a stable identifier meant for machines; `message` is meant for people reading logs.'
  license:
    name: MIT
    url: https://opensource.org/licenses/MIT
  version: 1.0.0
  contact:
    name: answerline support
    email: support@answerline.dev
  termsOfService: https://answerline.dev/legal/terms
servers:
- url: https://api.answerline.dev
  description: Production server
security:
- bearerAuth: []
tags:


# --- truncated at 32 KB (32 KB total) ---
# Full source: https://artifacts.apis.io/answerline-dev/openapi/answerline-dev-async-api-openapi.yml