> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gamecart.gg/llms.txt
> Use this file to discover all available pages before exploring further.

# Rate limits

> Per-token request limits, rate headers, and the executor cadence

Gamecart limits how often each API token can call each route. The limit counts per token, per HTTP method, and per route, so reading one order counts against `GET /v1/seller/orders/{orderId}` no matter which order you read.

| Method                           | Limit                   |
| -------------------------------- | ----------------------- |
| `GET`                            | 120 requests per minute |
| `POST`, `PUT`, `PATCH`, `DELETE` | 30 requests per minute  |

The window starts at the first request and lasts 60 seconds.

## Headers

Responses to authenticated requests carry the budget for that route:

```http theme={"dark"}
X-RateLimit-Limit: 120
X-RateLimit-Remaining: 117
```

When the budget runs out, the API answers `429` with the code `RATE_LIMIT_ERROR`, a `Retry-After` header in seconds, and `retryAfterSeconds` in the body:

```http theme={"dark"}
HTTP/1.1 429 Too Many Requests
Content-Type: application/problem+json
Retry-After: 17
X-RateLimit-Limit: 120
X-RateLimit-Remaining: 0
```

Wait at least `Retry-After` seconds before calling that route again. Other routes keep their own budget.

## Command execution

The three command execution routes, pull, done, and fail, do not count against the token limits above. They follow the cadence of the command executor protocol instead: 120 calls per minute for each server and each of the three routes.

That budget belongs to the server, not to the token. An official connector and your own executor working on the same server share it.

## Staying under the limits

* Use [webhooks](/webhooks/overview) to learn about new and changed orders instead of polling the order list.
* Filter lists on the server with `from`, `status`, and the other query parameters instead of paging through everything.
* Keep one token per system, so a busy integration does not spend the budget of another.