Scheduled tasks

Scheduled tasks are stored on the WECOMMANDER service, not in the operating system's scheduler. Tasks on an agent device run through its agent, even while your own computer is off; tasks on a saved SSH server run from your desktop app, which holds the SSH credentials.

Create a task

  1. Open Scheduled tasks in the dashboard (Workspace → Scheduled tasks) or in the desktop app.
  2. Name it, choose the target (an agent device; SSH servers are chosen in the desktop app) and enter the exact command.
  3. Choose the schedule, the time zone, the time limit and when to e-mail you. The next five run times are shown before you save.
  4. Confirm. Creating the task is the approval for its runs: it runs at each due time without asking again. Check the command before saving.

Schedules

KindExampleNotes
Every N minutesevery 30 minutes5 minutes to 7 days; the first run is one interval after saving.
Cron (advanced)0 3 * * * (03:00 every day)Standard five fields: minute hour day month weekday; ranges, steps, lists and month/weekday names. A day matches if either day field matches.
Once2026-12-01T09:00+03:00A time in the past is refused.

Times are wall-clock times in the task's IANA time zone (default Europe/Istanbul). A time skipped by a daylight-saving change runs right after it (02:30 becomes 03:30); a time that occurs twice runs once.

Options

Runs and history

Who can do what

Anyone in the workspace can read tasks; creating, changing, deleting and running them needs the owner or admin role. Every change is in the audit log with the exact command. The agent's own limits still apply: a device with commands switched off refuses the run.

Limits

The Free plan keeps up to 10 scheduled tasks; no workspace can have more than 100.

For AI

scheduled_tasks_list is read-only. scheduled_task_create (agent devices only), scheduled_task_update, scheduled_task_delete and scheduled_task_run_now need a human approval of that exact change, unless the task's device is on full trust. Tasks on SSH servers always need an approval, command allowlists do not apply, and the device's safety net refuses catastrophic commands.