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: Atlassian Compass
providerId: atlassian-compass
generated: '2026-09-06'
method: searched
source: https://developer.atlassian.com/cloud/compass/swagger.v3.json (operation descriptions and
  429 responses in Atlassian's published OpenAPI) and https://developer.atlassian.com/cloud/compass/rest/
created: '2026-05-04'
modified: '2026-09-06'
description: >
  Rate limits Atlassian actually publishes for Compass, read from the operation descriptions and
  429 responses in the Compass REST OpenAPI. This file replaces a 2026-05-04 scaffold whose tiers,
  quotas and headers were bulk-sweep defaults rather than provider-published values - in particular
  the scaffold advertised X-RateLimit-* and RateLimit-Policy response headers that Atlassian does
  not document for Compass.
tags:
- Atlassian
- Component Management
- Developer Experience
- Rate Limiting
limit_count: 3
scope_model: per authenticated user
headers:
  published: false
  note: >
    Atlassian publishes no rate-limit response headers for the Compass API. The OpenAPI declares a
    429 response with no headers block, and the REST reference documents no X-RateLimit-*,
    RateLimit-* or Retry-After semantics for Compass. An agent has the documented number but no
    runtime signal, and must treat a 429 as the only feedback.
  observed:
    probed: false
    reason: every Compass endpoint requires authentication, so no unauthenticated response was
      available to read headers from
responseCodes:
  throttled: 429
limits:
- name: Metric ingestion
  scope: per user
  applies_to:
    operationId: insertMetricValue
    path: /compass/v1/metrics
    method: post
  metric: requests_per_minute
  limit: 100
  timeFrame: minute
  burst: null
  tier: all
  evidence: >
    "This API is rate limited. Only 100 requests per user per minute are allowed." -
    paths./compass/v1/metrics.post.description
- name: Event ingestion
  scope: per user
  applies_to:
    operationId: createCompassEvent
    path: /compass/v1/events
    method: post
  metric: requests_per_minute
  limit: 100
  timeFrame: minute
  burst: null
  tier: all
  evidence: >
    "This API is rate limited. Only 100 requests per user per minute are allowed." -
    paths./compass/v1/events.post.description
- name: Incoming webhook invocation
  scope: per webhook
  applies_to:
    operationId: handleWebhookInvocation
    path: /compass/v1/webhooks/{webhookId}
    method: post
  metric: unspecified
  limit: null
  timeFrame: null
  tier: all
  evidence: >
    The operation declares a 429 response ("Returned if the request exceeds the rate limit.") but
    Atlassian publishes no numeric ceiling for it.
undocumented:
- surface: Atlassian platform GraphQL API (https://api.atlassian.com/graphql)
  note: No Compass-specific GraphQL rate limit is published on the Compass GraphQL reference.
- surface: Atlassian Rovo MCP Server (https://mcp.atlassian.com/v2/mcp)
  note: >
    No request-rate ceiling is published. Atlassian instead documents a Rovo credit cost for some
    tools - the Teamwork Graph context tool is noted as consuming up to 10 Rovo credits per call.