specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: Autoura
providerId: autoura
generated: '2026-09-13'
method: searched
source: https://www.autoura.com/docs/api
created: '2026-05-04'
modified: '2026-09-13'
supersedes:
  method: generated
  modified: '2026-05-04'
  note: >-
    The previous file was a bulk-sweep scaffold asserting X-RateLimit-* headers, a 429
    contract and per-minute limits of 10/100/1000. NONE of that is published by
    Autoura. It has been replaced with what the provider actually documents, which is
    nothing on request rate and three quotas on consumption.
tags: [Tourism, Travel, Quotas, Rate Limiting]
description: >-
  Autoura publishes NO request-rate limits and NO rate-limit response headers. What it
  does publish are plan-level consumption quotas, which throttle how much content and
  how many consumer profiles an account can create -- not how fast it can call.
limit_count: 0
limit_count_note: >-
  An honest zero. Every documentation surface was read -- the API reference, the
  authentication and concepts pages, the changelog, both agent skills and the
  authentication references -- and no per-second, per-minute, per-hour or per-day
  request limit appears on any of them.
headers:
  limit: null
  remaining: null
  reset: null
  retryAfter: null
  policy: null
headers_note: >-
  No X-RateLimit-*, RateLimit-* or Retry-After header is documented, and none was
  observed on the live unauthenticated 401 responses probed on 2026-09-13 against
  /api/whoami, /api/stops/search, /api/visits/search and /api/characters/search.
responseCodes:
  throttled: null
  quotaExceeded: null
  note: No 429 contract is published. The only status observed unauthenticated is 401.
limits: []
quotas:
  note: >-
    These are consumption ceilings attached to a subscription tier, taken from the
    published pricing page. They limit what an account can CREATE, not how often it
    can call. Full tier detail in plans/autoura-plans-pricing.yml.
  entries:
    - {tier: Starter, metric: new_profiles, limit: 150, timeFrame: month, escape_hatch: 'bring your own OpenAI/ElevenLabs key'}
    - {tier: Scale, metric: new_profiles, limit: 150, timeFrame: month, escape_hatch: 'bring your own OpenAI/ElevenLabs key'}
    - {tier: Enterprise, metric: new_profiles, limit: 150, timeFrame: month, escape_hatch: 'bring your own OpenAI/ElevenLabs key'}
    - {tier: Starter, metric: moveme_generations, limit: 10, timeFrame: month}
    - {tier: Scale, metric: moveme_generations, limit: 50, timeFrame: month}
    - {tier: Enterprise, metric: moveme_generations, limit: 100, timeFrame: month}
    - {tier: Starter, metric: characters, limit: 3, timeFrame: account}
    - {tier: Scale, metric: characters, limit: 8, timeFrame: account}
    - {tier: Starter, metric: visits, limit: 2, timeFrame: account}
    - {tier: Scale, metric: visits, limit: 10, timeFrame: account}
    - {tier: Enterprise, metric: visits, limit: 25, timeFrame: account}
    - {tier: Scale, metric: routes, limit: 10, timeFrame: account}
    - {tier: Enterprise, metric: routes, limit: 50, timeFrame: account}
policies:
  - name: Caching in place of throttling
    description: >-
      Autoura fronts the data and audio APIs with the Fastly CDN and images with
      Cloudinary. A cached read does not reach origin, which is plausibly why no
      request-rate limit is published -- but it gives a client no runtime signal
      either way.
    source: https://www.autoura.com/docs/api/concepts
  - name: Polling cadence (documented, not enforced)
    description: >-
      The only cadence Autoura publishes anywhere is guidance for its own
      preference-sharing flow - "In our integrations we poll every 750ms", with the
      QR code withdrawn after 180 seconds. That is advice, not a limit.
    source: https://www.autoura.com/docs/api/integrations/signin
gaps:
  - id: no-runtime-rate-signal
    severity: high
    detail: >-
      An agent calling Autoura has no way to learn how much budget it has left or when
      to back off. There is no header to read, no documented 429, and no published
      number to plan against. For a surface whose primary consumer is now an
      autonomous agent, this is the largest single ergonomic gap in the profile.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com