generated: '2026-08-12'
method: searched
source: https://docs.thebrief.ai/public-api
docs:
  - https://docs.thebrief.ai/public-api
  - https://docs.thebrief.ai/public-api/rest-api/error-codes
provider: The Brief
providerId: thebrief
description: >-
  The Brief publishes two per-team rate limits on the Public API, split between export
  routes and everything else. The numbers are documented; the runtime signal is not —
  no RateLimit-*, X-RateLimit-* or Retry-After response header is described anywhere in
  the documentation, so a client can only discover exhaustion by receiving a 429.
limit_count: 2

limits:
  - name: Export routes
    scope: team
    metric: requests
    limit: 15
    window: 10 seconds
    burst: null
    applies_to:
      - POST https://api.thebrief.ai/v1/export
      - POST https://api.thebrief.ai/v1/export-with-changes
    note: >-
      Documented verbatim as "Export: 15 requests / 10 seconds / team". Exporting a design SET
      counts as a single request, not one per size.
  - name: All other routes
    scope: team
    metric: requests
    limit: 100
    window: 10 seconds
    burst: null
    applies_to:
      - 'https://api.thebrief.ai/v1/* (non-export)'
    note: 'Documented verbatim as "Rest of the routes: 100 requests / 10 seconds / team".'

response:
  exhaustion_status: 429
  exhaustion_meaning: Too many requests.
  headers:
    documented: false
    checked:
      [RateLimit-Limit, RateLimit-Remaining, RateLimit-Reset, X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Reset, Retry-After]
    note: >-
      None of these are documented. This is the single biggest agent-readiness gap on The
      Brief's API: the limit is knowable at design time but not at runtime, so an autonomous
      client cannot pace itself and must treat 429 as the only feedback channel.
  retry_guidance:
    documented: false
    note: No documented backoff or retry policy; combined with the absence of idempotency keys, retries on export routes risk duplicate designs and duplicate credit spend.

graphql:
  endpoint: https://graphql.thebrief.ai/public
  documented_limits: false
  note: >-
    No separate GraphQL rate limit is published. The schema exposes an InputApiRateLimit input
    type (group, requestTimesPerDuration, cacheDuration), showing the platform models per-group
    request-times-per-duration limits internally, but no values are published.
  source: graphql/thebrief-public.graphql (input InputApiRateLimit)

quota_vs_rate:
  note: >-
    Distinct from rate limiting, API work consumes team CREDITS. Credit balance and per-operation
    credit cost are readable at GET /v1/credits/subscription-info and GET /v1/credits/operations.
    Ad-serving impressions are capped per plan (100,000/month on Ultra and Team, custom on Enterprise).
  see: plans/thebrief-plans-pricing.yml

cross_links:
  conventions: conventions/thebrief-conventions.yml
  errors: errors/thebrief-error-codes.yml
  plans: plans/thebrief-plans-pricing.yml