Rate Limits

LogirAI applies rate limits at three layers: public unauthenticated endpoints (per IP), authenticated APIs (per credential), and outbound webhook delivery (per subscriber).

Authenticated endpoints

There is no global per-credential rate limit in v1. Practical caps come from:

• Per-endpoint payload limits (10 MB JSON, 20 MB multipart for the pre-alert upload).
• Per-list pagination caps where applicable.

If you anticipate sustained throughput above ~50 requests / second from a single API key or company, contact LogirAI ops to coordinate so we can pre-warm capacity.

Public endpoints (per IP)

GET /health (liveness probe) is unlimited.

Outbound webhook delivery

LogirAI fires webhook events to your endpoint as they happen. Delivery characteristics in v1:

Property	Value
Retries	None — fire-and-forget, single attempt.
Timeout	10 seconds end-to-end per delivery. If your endpoint takes longer, the delivery is recorded as failed and the event is lost.

If you need durable delivery with retries and backoff, the v2 webhook system (with HMAC-SHA256 signatures, per-event subscription, dead-letter queue) is on the roadmap — see Changelog.

Best practices

• Cache list reads — guide listings change at the cadence of scans. Caching responses for 30–60 s on your side reduces load on both ends.
• Use webhooks instead of polling — once you've received guide.status_changed, you don't need to re-fetch unless reconciliation is required.
• Stagger bulk operations — when calling bulk-advance-status or bulk-customs-status, batch in chunks of ≤ 200 IDs and serialize calls.
• Honor Retry-After — when a 429 response includes the header, wait at least that many seconds before retrying.