Rate limits
Request limits for the platform API, MCP tools and connectors, the headers that report them and how to back off.
Updated 07/10/2026
Platform API limits
Limits apply per client IP address over a one-minute window. The first matching rule applies:
| Route | Limit per minute |
|---|---|
/mcp | 60 |
Sandbox API explorer (…/sandbox/test) | 30 |
| AI assistant routes | 20 |
| Billing and checkout | 30 |
| Webhook test and retry actions | 20 |
/api/developer-portal/me and /workspace | 120 |
| SCIM provisioning | 120 |
| Everything else | 300 |
Headers
Every response includes:
| Header | Meaning |
|---|---|
X-RateLimit-Limit | Requests allowed in the window |
X-RateLimit-Remaining | Requests left in the window |
X-RateLimit-Reset | When the window resets, in Unix seconds |
When you exceed a limit the API returns 429:
{
"error": "rate_limit_exceeded",
"message": "Too many requests. Please wait before retrying.",
"retryAfterSeconds": 12
}Wait retryAfterSeconds (or until X-RateLimit-Reset) before retrying, and add jitter if many clients share an address.
MCP tool limits
Each MCP tool also has a per-credential limit, between 30 and 120 calls a minute for platform tools and as reviewed for imported tools. See Use the Nexra MCP server.
Provider limits in Connect
Each connector page lists the provider's documented limits. When a provider responds with a rate limit, Nexra:
honours Retry-After (up to 20 seconds per wait);
retries reads and safely repeatable writes up to three times;
marks the run waiting_for_rate_limit while it waits.
Writes without duplicate protection and billable writes are never retried. If a flow regularly hits provider limits, reduce its schedule frequency or page size, or split it by entity.