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

# Rate limits

> Every response carries your quota — even the ones that reject your request

Every response — including `401` and `403` — carries these headers:

| Header                  | Meaning                                               |
| ----------------------- | ----------------------------------------------------- |
| `X-RateLimit-Limit`     | Requests allowed per window, for this key             |
| `X-RateLimit-Remaining` | Requests left in the current window                   |
| `X-RateLimit-Reset`     | Seconds until the window resets — **not** a timestamp |

## When you're over the limit

A `429` response adds a `Retry-After` header (seconds to wait) and the standard error body:

```json theme={null}
{
  "error": {
    "code": "rate_limited",
    "message": "Quota de requêtes dépassé — réessayez dans quelques secondes",
    "details": {
      "retry_after_seconds": 42
    }
  }
}
```

## Backing off

```bash theme={null}
response=$(curl -s -w "\n%{http_code}" "http://127.0.0.1:4010/automations?limit=1" \
  -H "Authorization: Bearer agk_live_your_key_here")
status=$(echo "$response" | tail -1)

if [ "$status" = "429" ]; then
  retry_after=$(echo "$response" | head -1 | jq -r '.error.details.retry_after_seconds')
  sleep "$retry_after"
  # retry
fi
```

Reading `X-RateLimit-Remaining` on every response and slowing down before you hit `0` is friendlier than waiting for a `429` — and every response gives you that number for free.

<Note>
  Repeated authentication failures from the same address can also trigger `429`, independently of your per-key quota.
</Note>