Scenario: Scheduled Tasks & Automation Workflows
When you schedule recurring tasks, the agent executes them at the specified times or intervals. Start with everyday jobs like morning news summaries, weekly reports, and regular file cleanup.
🎯 Objectives
- Automate daily and weekly recurring tasks with a single setup.
- Generate reports, summaries, and maintenance jobs without manual checks.
- Optionally build multi-agent automation workflows across sessions.
📋 Step-by-Step Guide
Step 1: Define Tasks to Automate
Identify recurring work clearly. Start with everyday examples; treat Git and server checks as optional.
Everyday — daily examples:
- Every day at 09:00: Summarize top 10 AI news articles.
- Every day at 18:00: Summarize incoming emails received today.
- Every day at 00:00: Clean up the downloads folder.Everyday — weekly examples:
- Every Monday at 10:00: Generate the weekly project report.
- Every Friday at 17:00: Summarize meeting notes from this week.Everyday — monthly examples:
- On the 1st of every month: Generate the monthly sales report.
- On the last Friday of every month: Organize next month's to-do list.Optional (developers): Git commit digests and server health pings appear later only if you need them.
Step 2: Schedule the Work (Scheduled Tasks UI recommended)
The easiest path is sidebar Scheduled Tasks → create a task with a schedule and message. You can also ask in chat.
Basic schedule request (everyday):
Schedule the following task every day at 09:00:
- Search for 10 recent AI news articles.
- Generate a one-line summary for each article.
- Include source links.
- Save results to 'daily-reports/ai-news-{date}.md'.Complex schedule request (everyday):
Run the following workflow every Friday at 17:00:
1. Find PDFs received this week in ./downloads/.
2. Summarize each file in one line.
3. Save the summary to 'weekly-reports/files-{date}.md'.Optional (developers) — weekly Git summary:
Every Monday at 10:00, summarize last week's Git commits
and save to 'weekly-reports/weekly-{date}.md'.Step 3: Inspect and Update Schedules
Review active tasks anytime in sidebar Scheduled Tasks, or ask in chat.
List active tasks:
Show all currently scheduled automation tasks.
Summarize the execution time and content for each task.Modify a schedule:
Change the execution time of the daily 09:00 AI news report to 14:00.Pause or remove a schedule:
Pause the weekly Friday file-summary schedule temporarily.Step 4: Verify Task Execution
Inspect the outputs generated by the automated agent.
Request verification:
Check all automated reports generated during the past week.
Summarize which reports were created.Check error reports:
Check if any automated tasks failed recently.
If a task failed, show the root cause and the recommended fix.💡 Example Prompt Collection
| Purpose | Prompt |
|---|---|
| Daily news summary | "Summarize top 10 AI news articles daily at 09:00 with source links." |
| File cleanup | "Find and remove duplicate files in ./downloads/ every Friday at 17:00." |
| Email summary | "Summarize all incoming emails daily at 08:00." |
| Data backup | "Create a backup of ./data/ into ./backups/ every midnight." |
| Learning report | "Generate a summary report of topics learned this week every Sunday at 18:00." |
| Optional: weekly Git report | "Generate a weekly Git commit analysis report every Monday at 10:00." |
| Optional: review reminder | "Send a reminder for pending code reviews every Wednesday at 15:00." |
| Optional: system monitoring | "Check system resource usage hourly and notify if it exceeds 80%." |
⏰ Schedule Frequency Guide
Daily Execution
Best suited for:
- News summaries and daily status reports.
- Routine file cleanup and local backups.
- Email and inbox notifications triage.
Example prompt:
Execute the following task every day at 09:00:
1. Search for 10 recent AI news articles.
2. Provide a one-line summary for each article.
3. Save results to daily-reports/{date}.md.Weekly Execution
Best suited for:
- Weekly status reports and meeting notes.
- Folder cleanup.
- Weekly metric summaries.
Everyday example:
Run this task every Friday at 17:00:
1. Clean ./downloads/.
2. One-line summaries for new PDFs.
3. Save to weekly-reports/.Optional (developers):
Run this task every Monday at 10:00:
1. Review Git commits from last week.
2. Summarize key modifications.
3. Save the report to weekly-reports/.Monthly Execution
Best suited for:
- Monthly sales and performance reviews.
- To-do and calendar cleanup.
Example prompt:
Run this task on the 1st of every month at 10:00:
1. Scan last month's notes and report folders.
2. Generate the monthly overview report.Interval Execution — Optional
Infrastructure checks (server ping, error-log spikes) are optional.
Check the following items every hour at minute 0:
- Web service response latency.
- Recent error log spikes.
- Send an alert if anomalies are detected.⚙️ Configuration Tips
Create a Dedicated Automation Assistant
Use a dedicated agent profile for automated tasks:
- Open Assistants in the sidebar and select New Assistant.
- Name:
"Automation Assistant"or"Scheduled Agent". - System prompt:
You are an automated task execution assistant. Execute scheduled jobs accurately and quickly. Record results clearly and report root causes upon failure. - Enable only the required tools.
Configure Failure Policies
Define error handling behavior explicitly:
When running automated tasks:
- Save error details to 'errors/' if a failure occurs.
- Provide a summary of failed tasks every morning.
- Send an urgent notification if critical tasks fail.Resource Management
Prevent excessive concurrent background tasks:
When setting up automated tasks:
- Distribute schedules across different time slots.
- Schedule heavy CPU or memory jobs during off-peak hours.
- Limit concurrent active tasks to a maximum of 3.📊 Automation Dashboard
You can ask the agent to summarize all active automation jobs in a structured dashboard:
Show all configured automation tasks in a dashboard table:
- Task Name
- Schedule Interval
- Last Run Time
- Recent Failure Status
- Next Scheduled Run🔗 Related Documentation
- Playbooks Guide — Save repeatable workflows into playbooks
- Assistants Guide — Configure dedicated automation agents
- Deep Research Scenario — Connect automation to scheduled research pipelines
🚀 Next Steps
| Project | Description | Difficulty |
|---|---|---|
| Daily News Digest | Summarize AI news every morning | ⭐⭐ |
| Automated Cleanup | Clean downloads directory on schedule | ⭐⭐ |
| Email Summary | Summarize daily morning emails | ⭐⭐ |
| Automated Backup | Configure periodic data directory backups | ⭐⭐ |
| Optional: Weekly Git Report | Generate weekly commit analytics | ⭐⭐⭐ |
| Multi-Agent Pipeline | Connect multiple agents into one pipeline | ⭐⭐⭐⭐ |
| Failure Notifications | Receive instant alerts on job errors | ⭐⭐ |
| Automation Monitor | Build a status dashboard for tasks | ⭐⭐⭐ |