Create and manage Temporal Schedules for recurring workflows via SDK and the temporal CLI

Documented steps

1. Create a schedule with the CLI: `temporal schedule create --schedule-id "YourScheduleId" --calendar '{"dayOfWeek":"Fri","hour":"3","minute":"30"}' --workflow-id YourBaseWorkflowId --task-queue YourTaskQueue --type YourWorkflowType`. Alternatively use `--cron` for Unix cron syntax or `--interval` for shorthand durations such as `10s`.
2. Create a schedule from Python: `await client.create_schedule(schedule_id, Schedule(action=ScheduleActionStartWorkflow(MyWorkflow.run, id=..., task_queue=...), spec=ScheduleSpec(intervals=[ScheduleIntervalSpec(every=timedelta(minutes=10))])))`.
3. Create a schedule from TypeScript: `await client.schedule.create({ scheduleId, action: { type: 'startWorkflow', workflowType, args, taskQueue }, spec: { intervals: [{ every: '10s' }] }, policies: { catchupWindow, overlap } })`.
4. Get a handle for subsequent operations: Python `handle = client.get_schedule_handle(schedule_id)`; TypeScript `const handle = client.schedule.getHandle('schedule-id')`. The CLI operates directly against `--schedule-id`.
5. Pause and unpause: SDK `await handle.pause(note="...")` / `await handle.unpause(...)`; CLI `temporal schedule toggle --schedule-id ID --pause --reason "..."` or `--unpause --reason "..."`.
6. Trigger an out-of-band run: SDK `await handle.trigger()`; CLI `temporal schedule trigger --schedule-id ID --overlap-policy ...`. Triggered runs are still subject to the schedule's Overlap Policy.
7. Backfill missed runs over a time window: Python `await handle.backfill(ScheduleBackfill(start_at=..., end_at=..., overlap=ScheduleOverlapPolicy.ALLOW_ALL))`; TypeScript `handle.backfill({...})`; CLI `temporal schedule backfill --schedule-id ID --start-time ... --end-time ... --overlap-policy BufferAll`.
8. Choose an Overlap Policy at create or update time: `Skip` (default), `BufferOne`, `BufferAll`, `CancelOther`, `TerminateOther`, or `AllowAll` — `AllowAll` is the only policy that permits truly concurrent runs.
9. Update a schedule: prefer the SDK read-modify-write form `await handle.update(callback)` which receives a `ScheduleUpdateInput` and returns a `ScheduleUpdate`. The CLI `temporal schedule update` uses full-replacement semantics.
10. Inspect: `handle.describe()` in the SDK, or `temporal schedule describe --schedule-id ID` and `temporal schedule list [--long|--really-long] [--query ...]`. Preview upcoming fire times without executing anything via `temporal schedule list-matching-times` (marked Experimental).
11. Delete when finished: `await handle.delete()` or `temporal schedule delete --schedule-id ID`. This removes only the schedule entity; workflow executions it already started are unaffected.
12. Inside the scheduled workflow, read `LastCompletionResult` / `LastFailure` to chain state between runs, and use the auto-injected search attributes `TemporalScheduledStartTime` and `TemporalScheduledById` for observability.
13. Reference docs: https://docs.temporal.io/schedule | https://docs.temporal.io/cli/schedule | https://docs.temporal.io/develop/python/schedules | https://docs.temporal.io/develop/typescript/schedules