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: Facets
providerId: facets
generated: '2026-09-07'
method: searched
source: >-
  https://www.facets.cloud/docs (full docs sitemap of 282 URLs walked),
  https://www.facets.cloud/docs/api, https://www.facets.cloud/pricing, and a scan of all 629
  operations in openapi/_original/facets-control-plane-openapi.json.
created: '2026-05-04'
modified: '2026-09-07'
limit_count: 0
note: >-
  REPLACES A SCAFFOLD, AND THE HONEST ANSWER IS ZERO. Until 2026-09-07 this file asserted
  invented limits (10 req/min free, 100 req/min professional, 429 on exhaustion, a full set
  of X-RateLimit-* headers) written by the 2026-05-04 bulk sweep. None of it is published by
  Facets. What we actually found:
    - The OpenAPI document contains zero occurrences of "RateLimit", "X-RateLimit" or
      "Retry-After", and declares no 429 response on any of its 629 operations. The only
      statuses any operation declares are 200, 400, 403, 404, 405, 409 and 500.
    - No page in the 282-URL docs sitemap documents an API rate limit, quota or throttle.
    - Pricing (https://www.facets.cloud/pricing) meters Resource Instances, not API calls;
      there is no request quota in any tier.
  This is plausible rather than surprising for the deployment model: each customer gets their
  own control plane (SaaS) or self-hosts it entirely (Enterprise), so there is no shared
  multi-tenant API surface to protect with a published global limit. That explains the
  absence; it does not remove the consequence, which is that an automation client has NO
  published signal telling it when to back off, and no documented status code to key retry
  logic on.
headers: {}
headers_observed: []
responseCodes: {}
limits: []
policies: []
related_quotas:
  - name: Praxis per-user AI credit allocation
    scope: per-user, within an org-wide monthly budget
    window: billing cycle
    enforcement: 'Praxis blocks further usage once a user hits their cap, until the next cycle or a higher allocation from an admin.'
    source: https://www.facets.cloud/docs/changelog/release-notes-v092
    note: >-
      This is an AI SPEND cap on the Praxis agent layer, not an HTTP rate limit on the
      Control Plane API. Recorded here because it is the only published usage ceiling of any
      kind, and deliberately kept out of limits[] so it is not counted as an API rate limit.
recommendation: >-
  With nothing published, treat the API as unmetered but fragile: back off exponentially with
  jitter on 500, and re-read state rather than blind-retrying on 409, because the API has no
  idempotency mechanism (see conventions/facets-conventions.yml).
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com