# `SuperPi.Workers.ScheduleRunWorker`
[🔗](https://github.com/oleksandr/super-pi-beam/blob/v0.1.0/lib/super_pi/workers/schedule_workers.ex#L22)

Starts one fire of a schedule. Unique on `[:schedule_id, :fire_at]` forever (`period:
:infinity`), where `fire_at` is the scheduled minute (or a webhook's `Idempotency-Key`), so a
duplicate tick or a webhook retry inserts nothing. `SuperPi.Schedule.run/2` then claims the
fire with a conditional update, which makes an Oban retry or a `Lifeline` rescue harmless.

The job ends when the run task is stored and its conversation machine is told. It does not
wait for the turn, so it never holds a queue slot for the length of an LLM run. After the last
attempt a failure counts against the schedule like a failed run.

---

*Consult [api-reference.md](api-reference.md) for complete listing*