Cron
Three tools, cron_jobs, cron_runs and cron_control, let the agent list deployed scheduled jobs, read their run history, run a job now, pause it or resume its schedule.
A scheduled job is a file app/<folder>/<name>.cron.ts with export const schedule (UTC) and a default function. Deploy registers the schedule; a failed run is retried up to 3 times.
cron_jobs
Lists the deployed jobs.
- Parameters:
project;env:previeworprodto filter; both by default. - Returns: each job’s
id, schedule (UTC),paused, next run and last run. No jobs returns an empty list with a note. - When the agent calls it: after a deploy, to confirm the schedule was registered and to get the
job_id.
cron_runs
The run history of one job, newest first.
- Parameters:
project;job_id;limit: 1–200, default 20. - Returns: each run’s status, duration and error, if any.
- When the agent calls it: when a job doesn’t do what it should, or after
cron_controlwithrun.
cron_control
Controls one job.
- Parameters:
project;job_id;action:run: runs the job once now, even while paused. Returns arun_id; about 10 seconds later the result is incron_runs.pause: stops scheduled runs.resume: restarts them.
- Returns:
run_idforrun;pausedforpauseandresume. - When the agent calls it: to test a newly deployed job without waiting for its time, or to stop a job that sends the wrong emails.
A pause is kept across deploys. Each run executes the app’s real code, which may send email or call outside services.
Next steps
- Scheduled jobs: writing a
.cron.tsfile. - Logs: cron runs show up in
traces. - Plan limits: cron jobs per plan.
Last updated