Skip to main content

Scheduled tasks

Run a prompt on a schedule: "run the tests every weekday at 9", "summarize new files every Monday", "check the deploy hourly". Each run is a real chat in the workspace you choose, so you can read exactly what happened.

Create a task​

The fastest way is to ask. In any chat, in the workspace where the work should happen:

Schedule a task that runs the test suite every weekday at 9 and reports failures with the error output.

The agent has a tool for this; approve the tool call and the task exists, workspace and model already set from the chat. Then open the Scheduled page to check the next-run time and click Run now once before trusting the schedule.

The Scheduled page is also where you create and manage tasks by hand: name, prompt, model, workspace, and a schedule from the frequency builder (hourly, daily, weekly, ...) or a raw RRULE. Schedules are iCalendar RRULEs; the raw field takes anything RRULE supports:

ScheduleRRULE
Every day at 09:00RRULE:FREQ=DAILY;BYHOUR=9;BYMINUTE=0
Every MondayRRULE:FREQ=WEEKLY;BYDAY=MO
Every hourRRULE:FREQ=HOURLY;INTERVAL=1
Once, then never againRRULE:FREQ=DAILY;COUNT=1
Runs are unattended

Scheduled runs execute with full tool approval: nobody is there to click Allow. Give the task a workspace you're comfortable with and a prompt that says exactly what to do (and what not to).

What a run produces​

Every run creates:

  • a real chat in the selected workspace, visible in the sidebar with the full task activity, and
  • a run-history entry on the task showing success or error (with the error text).

Want a ping? Add a notification target once, then just put the condition in the task's prompt ("notify me only if it fails"): the agent has a notify tool and decides when to use it. Blanket event pings on every finished/failed run exist too. See notifications and webhooks.

Trigger a task with a webhook​

Any scheduled task can also be started from outside (CI, a cron job, another service) without a browser session:

  1. On the task, generate its webhook URL. It contains a secret token.
  2. POST to that URL to run the task immediately.
  3. Revoke or regenerate the URL anytime if it leaks.

If the POST has a JSON body, the prompt can use it via the {{webhook_payload}} placeholder, so a CI failure payload, form submission, or alert can flow straight into the task's instructions.

A good first task​

Name: Morning test check. Schedule: RRULE:FREQ=DAILY;BYHOUR=9;BYMINUTE=0. Prompt:

Run the test suite in this workspace. Don't change any files.
Report pass or fail; if it failed, include the failing test and error output.

Click Run now, read the chat it creates, and leave it enabled once the output is something you can judge in a few seconds.

This content is for informational purposes only and does not constitute a warranty, guarantee, or contractual commitment. Open WebUI is provided "as is." See your license for applicable terms.