generated: '2026-07-21'
method: derived
source: openapi/tinyfish-main-openapi.json (+ search/fetch/browser specs), https://docs.tinyfish.ai
authentication:
  style: api-key
  header: X-API-Key
  ref: authentication/tinyfish-authentication.yml
versioning:
  scheme: uri-path
  current: v1
  example: https://agent.tinyfish.ai/v1/automation/run
pagination:
  style: cursor
  request_params: [cursor, limit]
  limit_default: 20
  limit_max: 100
  response_fields: [data, pagination.total, pagination.next_cursor, pagination.has_more]
  notes: Opaque base64 cursor from the previous response; next_cursor is null when no more results.
idempotency:
  supported: false
  detail: >-
    Run-creation endpoints (/v1/automation/run, /run-async, /run-sse, /run-batch)
    explicitly do NOT support idempotency keys — retrying a failed request may
    create duplicate runs. Cancel/get batch endpoints are id-based and idempotent
    (calling twice returns consistent results), but there is no client-supplied
    idempotency-key contract for writes.
request_tracing:
  field: request_id
  location: response body (and error envelope)
  notes: Correlate support requests by request_id.
rate_limiting:
  signal: HTTP 429 with error_code RATE_LIMIT_EXCEEDED
  retry_after: honor Retry-After header when present
  ref: errors/tinyfish-error-codes.yml
error_envelope:
  content_type: application/json
  fields: [error_code, message, request_id]
  format: provider-json (not RFC 9457)
  ref: errors/tinyfish-error-codes.yml
webhooks:
  ref: asyncapi/tinyfish-agent-webhooks.yml
  events: [run.completed, run.failed, run.cancelled]
lifecycle:
  ref: lifecycle/tinyfish-lifecycle.yml