Scheduled agent run skipped: no webhook arrives, read the 200 receipt

If you start a Scheduled agent run while the previous run is still active, the default on_active_run of skip returns 200 with status skipped and skip_reason previous_run_active. A skipped run delivers no webhook, so code that waits for action.run.terminal will wait forever. Read the status in the response you already have.

What does each start outcome look like?

The API trigger reference separates the cases.

Why does the default surprise people?

Skip is the default, so a naive caller sees a 2xx and assumes a run is coming. Check the status field on every start response. Only a started run has a terminal event to wait for. Canceled runs also send no webhook, so the same rule holds: poll the receipt at GET /v1/action-runs/{id}.

What should my caller do?

Handle the response first, then decide whether to wait.

• If status is skipped, log it and stop. Do not register a waiter.
• If you want a hard failure instead, send on_active_run reject and treat the 409 as the alert.
• If a run did start, verify the webhook signature over timestamp.raw_body with HMAC-SHA256, refuse an empty secret, and dedupe on the envelope request_id, which equals run_id.
• Keep a poll as a backstop for the cases that send no event.

What is the input here?

Whatever you pass as input is data the agent reads, not instructions it obeys. A skip decision never reads it either. For the signature rules, retries (10 attempts) and the 5-minute replay window, see the run webhooks guide and the API trigger reference.