Skip to main content

Cron Jobs

Use cron jobs when a PUDA task should run on a schedule, such as every morning before the lab opens, once per hour during a long experiment, or every night after instruments are idle.

Cron is best for already-tested procedures. Before scheduling a machine task, turn the task into a protocol or workflow skill, dry-run it, and confirm that it handles startup, verification, cleanup, and failure cases safely.

What to schedule​

Schedule deterministic commands or something that has been done before, not vague instructions. Good cron targets include:

  • Running a saved protocol.
  • Asking an agent to execute a named workflow skill from a project directory.
  • Collecting status or telemetry from machines.
  • Generating reports at the end of the day.
  • Running routine startup, calibration, or shutdown tasks that have already been validated.

Avoid scheduling experimental commands that still need human judgement. If a task can move hardware, heat a device, dispense liquid, or consume a sample, the scheduled command should include explicit limits and verification gates.

Examples​

Describe the schedule in natural language and let the agent turn it into the right cron job for your project:

Do a morning startup check every day at 8:00 AM.
Do a weekly calibration every Monday at 9:30 AM on the Opentrons machine.
Collect machine status every hour during the long-running experiment.

Safety checks​

Before enabling the cron job:

  • Confirm the machine is reachable with puda machine list.
  • Use natural langauge to run at least once before turning it into a cronjob.
  • Write cron output to a log file so failures are visible later.

Cron will keep running on its schedule. Disable or edit the cron entry whenever the machine setup, protocol file, consumables, or safety limits change.