generated: '2026-08-13'
method: searched
source: https://demodesk.com/api/docs/index.html
note: >-
  Demodesk publishes rate limits inside the v2 OpenAPI itself — every operation's
  `description` carries a "Rate limits:" block. That makes the numbers
  provider-published rather than derived. The limits are enforced per API key.
  The 429 response declares a `Retry-After` header, so an agent is told how long
  to wait — but no RateLimit-* / X-RateLimit-* headers are declared, so remaining
  quota cannot be read before exhaustion, only after it.
limit_count: 2
limits:
- id: global-per-key
  scope: per-key
  limit: 120
  unit: requests
  window: 1m
  applies_to: all v2 operations
  source: openapi/_original/demodesk-v2-openapi.yml (operation descriptions)
  evidence: 'Global: 120 requests per minute per API key.'
- id: list-recordings-per-key
  scope: per-endpoint
  limit: 60
  unit: requests
  window: 1m
  applies_to:
  - listRecordings
  source: openapi/_original/demodesk-v2-openapi.yml#listRecordings
  evidence: 'Endpoint-specific: 60 requests per minute per API key.'
exhaustion:
  status: 429
  body_media_type: application/json
  body_shape: '{ "error": { "code", "message", "requestId" } }'
  ref: errors/demodesk-problem-types.yml
response_headers:
  documented: true
  retry_after: true
  headers:
  - name: Retry-After
    type: integer
    description: Seconds until the next request is allowed.
    returned_on: 429
    source: openapi/_original/demodesk-v2-openapi.yml#/components/responses/too-many-requests
  standard_headers: []
  note: >-
    The published `too-many-requests` response component declares exactly one
    header, `Retry-After` (integer seconds). No RateLimit-* (RFC 9331 draft) or
    X-RateLimit-* headers are declared anywhere in the spec, and the 200
    responses carry no `headers:` block at all — so a client learns its budget
    only by exhausting it. Recorded as a partial runtime signal: back-off is
    machine-readable, remaining quota is not.
burst:
  documented: false