Skip to main content
This guide puts an installed agent on a recurring schedule, verifies it, promotes it to live, and later clears it. You need the agent’s installation id, from jstm agent installations. All times are UTC.
1

Find the installation

The table includes Schedule and Next fire columns, so you can see which installations are already scheduled. Copy the installation id you want.
2

Set a schedule, starting in dry_run

Start in dry_run so nothing side-effecting happens while you confirm the cadence:
--cron takes a standard five-field cron expression, evaluated in UTC. --env defaults to dry_run, so you can omit it here.The expression is checked before anything is sent. A typo like --cron "not-a-cron" fails immediately with a non-zero exit and no change made.
3

Verify

Next fire tells you exactly when the first run will happen. Last fired stays — until it runs once. Add --json for machine-readable output.
4

Promote to live

When you’re happy with the dry runs, switch the run mode to live:
Always repeat both --cron and --env live. set sends the run mode every time and defaults it to dry_run, so running set again with only --cron silently drops the schedule back to dry_run.
5

Clear the schedule

To stop scheduled runs while leaving the agent installed:

​
Common cron expressions

All expressions are evaluated in UTC.

​
Notes

  • UTC only. There is no per-schedule timezone yet.
  • If the platform is down across several scheduled times, it fires the most recent missed one once on recovery, not a burst.
  • Changing the schedule resets the clock to the next future time. It won’t back-fire.

​
See also