generated: '2026-08-13'
method: searched
source: https://cogny.com/docs/api-overview-authentication
docs: https://cogny.com/docs/api-overview-authentication
also: https://cogny.com/pricing
limit_count: 4
checked: '2026-08-13'
summary: >-
  Cogny documents three named REST rate-limit tiers on its API overview page and
  meters the MCP surface separately, in tool calls per month, on the pricing page.
  The documented tiers are numbers in prose only: Cogny publishes NO rate-limit
  response headers, so an agent has no runtime signal and can only discover
  exhaustion from the 429 status itself.
limits:
- id: rest-standard
  surface: rest
  scope: per-account
  tier: Standard
  window: 1m
  limit: 100
  unit: requests
  burst: null
  source: https://cogny.com/docs/api-overview-authentication
- id: rest-enterprise
  surface: rest
  scope: per-account
  tier: Enterprise
  window: 1m
  limit: 1000
  unit: requests
  burst: null
  source: https://cogny.com/docs/api-overview-authentication
- id: rest-batch
  surface: rest
  scope: per-endpoint
  tier: Batch operations
  window: 1m
  limit: 10
  unit: requests
  burst: null
  note: Applies to batch endpoints (e.g. bulkUpdateTickets, exportTickets).
  source: https://cogny.com/docs/api-overview-authentication
- id: mcp-free-tier
  surface: mcp
  scope: per-workspace
  tier: Free
  window: 1mo
  limit: 15
  unit: tool-calls
  burst: null
  note: >-
    Metered MCP tool calls on the free tier, reset at the start of every month.
    The Solo tier ($9/mo) is unmetered "under fair use" — no numeric ceiling is
    published for it. Cloud ($499/mo) meters in 2000 AI credits per month across
    chat, MCP calls and image generation.
  source: https://cogny.com/pricing
exemptions:
- subject: sk_test_ development keys
  note: >-
    "Development keys access sandbox data and don't count toward rate limits."
  source: https://cogny.com/docs/api-overview-authentication
response_signaling:
  headers_published: false
  headers: []
  ratelimit_header_standard: null
  retry_after: null
  status_on_exhaustion: 429
  status_meaning: 429 Too Many Requests - Rate limit exceeded
  error_envelope: '{ success: false, error: { code, message, details } }'
  note: >-
    GAP. The API overview documents 429 as the exhaustion status and documents a
    custom error envelope, but names no X-RateLimit-*, RateLimit-* (RFC 9331) or
    Retry-After header. Nothing in openapi/ declares rate-limit response headers
    either. An agent therefore cannot pace itself; it can only back off after
    being refused. This is the single highest-value runtime-semantics fix
    available to Cogny.
  see: conventions/cogny-conventions.yml
live_observation:
  attempted: true
  url: https://api.cogny.com/v1/reports
  http_status: 521
  checked: '2026-08-13'
  note: >-
    Headers could not be observed live: the documented REST base host
    api.cogny.com returned Cloudflare 521 (origin unreachable) on every probe on
    the checked date, across three retries and every path tried. Recorded as a
    probe result, not as a claim about the limits themselves — the numbers above
    come from the published docs.