Get started
Rate Limits
Endpoints are grouped into tiers. All endpoints in the same tier share a single per-user bucket.
| Tier | Limit | Endpoints |
|---|---|---|
| Standard | 300 req / min | Read endpoints (users, audit logs, API keys, agents, projects, budgets, workflows, file search and status, parse status), plus stop and delete agent |
| Write | 60 req / min | Mutations (disable/enable user, revoke API key, create/update projects and members, budget member limits, create/update/release workflows) |
| Heavy | 30 req / min | Resource-intensive (launch an agent, agent follow-up, artifact export, file upload, submit parse) |
| Search | 60 req / min | Search endpoints (codes search) |
| Analytics | 30 req / min | Organization-wide analytics endpoints, except organization agent sessions |
| Session analytics | 20 req / min | Organization agent sessions (GET /analytics/agents) |
Launching agents shares the Heavy bucket with uploads and parse submissions, so a job that uploads a file and then launches an agent consumes two requests from the same 30 / min budget.
Every successful response includes rate-limit headers:
| Header | Description |
|---|---|
X-RateLimit-Limit | Maximum requests allowed per window |
X-RateLimit-Remaining | Approximate remaining requests |
X-RateLimit-Reset | Unix timestamp (seconds) when the quota resets |
When a limit is exceeded the API returns 429 Too Many Requests with a Retry-After header (seconds). Clients that sleep for this duration will be paced at a steady rate.