wait-looplisted
Install: claude install-skill wenxueru/wait-skill
# Wait Loop
Run a task repeatedly on a timer.
Manage durable loop state through `../scripts/waitctl.py loop -- ...`; the service registers its timer watchers using the parent `wait` implementation. The service delegates loop validation and persistence to `wait_loop.py`. Read [the loop protocol](../docs/wait-loop.md), [the supervisor guide](../docs/waitd.md), [the watcher protocol](../docs/wait.md), and [the client adapters](../docs/clients.md) before starting.
## Invocation
- `<invoke> <interval>: <task>` starts a loop and runs the first iteration immediately.
- `<invoke> resume <state-file>` handles a timer event.
- `<invoke> status <state-file>` reports state without advancing it.
- `<invoke> cancel <state-file>` cancels future iterations.
Here, `<invoke>` is `$wait-loop` or `/wait-loop` according to the client.
## Protocol
1. Run `waitctl.py loop -- init` with the task, interval, client, session, and a finite overall `--duration`. Retain its returned `state_file`. The 24-hour default is a safety limit, not a reason to omit a more meaningful bound. Add `--max-iterations` when the requested count is finite.
2. Execute the saved task once. The root session performs each iteration; do not create an autonomous child loop.
3. After the iteration succeeds, run `complete --summary`. If it returns `completed`, report the final outcome and stop. Otherwise retain its `watch_id` and `next_run_at`.
4. The service registers the timer before returning from `complete`. Use the save