Skip to main content

Scheduled Automation

Scenario

You’ve built a daily briefing (or any repeatable task) and you want it to run automatically every morning at 9 AM, with the result pushed to your WeChat, Slack, or email — no manual trigger needed.

Materials

  • A working task or Skill you want to automate (e.g., the Daily Briefing)
  • A delivery channel configured in Settings → Channels (optional — results also appear in the task dashboard)

Steps

1

Describe Your Schedule in Chat

In any conversation, tell the agent:
“Every weekday at 9 AM, run my ‘Morning Briefing’ Skill and send the result to my WeChat.”
The agent detects the scheduling intent and presents a confirmation card showing the task name, schedule (cron expression), bound Skill, and delivery channel.
2

Confirm the Task

Review the confirmation card and click Confirm. The task is created instantly. A task card appears with a link to the management dashboard.
3

Monitor in the Dashboard

Go to Settings → Scheduled Tasks to see all your recurring tasks. Each entry shows the schedule, last run time, status, and output history.
4

Adjust or Pause

Click any task to edit the schedule, change the delivery channel, update the prompt, or pause execution. You can also manage tasks via chat:
“Pause my morning briefing” or “Change it to run at 8:30 AM instead”
Three task types: Agent tasks (full LLM execution), Shell commands (local/desktop only), and Python scripts (zero LLM cost, works everywhere including cloud). See Scheduled Tasks for the full reference.

Deliverable

An automated recurring task that runs on your defined schedule and delivers results to your chosen channel without manual intervention.

Acceptance Criteria

  • Task runs at the scheduled time
  • Output is delivered to the configured channel
  • Task appears in the Scheduled Tasks dashboard
  • You can pause, edit, or delete the task

What’s Next

  • Add reliability checks → Set up three successful manual runs before automating (a best practice for production workflows)
  • Chain tasks → Use task chaining to run dependent tasks in sequence (e.g., data collection → analysis → report)