Automations
Schedule, webhook, or API-triggered runs by an agent or a squad — with or without a traceable issue.
An automation (Autopilot) runs unattended work on a trigger: a schedule, a webhook, or an API call. The executor is an agent or a squad. It is how recurring or event-driven work stops depending on someone remembering to start it.
Triggers
| Trigger | When it fires |
|---|---|
| Schedule | On a time schedule — daily sweeps, weekly reports, periodic checks |
| Webhook | When an external system calls the automation's webhook |
| API | When your own tooling calls the API |
| Manual | Whenever you press run |
What a run does
Each run can create a traceable issue — the recurring work then has a status, comments, and history like any other work, and people can follow it in their inbox. Or a run can execute without creating an issue — right for runs that only matter when they find something (a nightly check that reports only on failures).
Operating an automation
The automation page gives you the controls to keep unattended work trustworthy:
- Manual run — trigger a run on demand without waiting for the schedule.
- Run history — every past run, so success is a fact rather than an assumption.
- Webhook delivery records — see the webhook calls that came in and how they were handled.
- Replay — re-run a past execution when a run failed or you need to reproduce it.
- Access management — control who can manage the automation.