Concept

Webhook-Triggered LLM Work

Webhook-triggered LLM work starts when your handler receives an event and needs a model request to complete safely.

The ReqRun boundary

ReqRun does not receive, verify, sign, or route callbacks in v1. Use ReqRun from your webhook handler when the event triggers an OpenAI request that needs retries and status.

How ReqRun helps

Use the provider event id as the idempotency key. If the sender retries the webhook, ReqRun returns the same durable request instead of creating duplicate LLM work.