pi-dagu-workflow Extension

A comprehensive Pi extension for Dagu workflow orchestration. Create, validate, deploy, and manage Dagu workflows directly from Pi.

Features

🛠️ Custom Tools

• dagu_create_workflow - Create workflows from natural language requirements
• dagu_validate_workflow - Validate workflow YAML files
• dagu_run_workflow - Execute workflows manually
• dagu_get_status - Check workflow status and logs
• dagu_template_workflow - Create from predefined templates
• dagu_manage_scheduler - Start/stop/restart the Dagu scheduler

⌨️ Commands

• /dagu-install - Install Dagu on macOS/Linux
• /dagu-start - Start the scheduler
• /dagu-stop - Stop the scheduler
• /dagu-list - List all workflows
• /dagu-status [workflow] - Check workflow status

📋 Templates

• etl_pipeline - Data extraction, transform, load
• ci_cd - Build, test, deploy pipeline
• backup_job - Automated backups with verification
• api_monitor - API health checking
• data_processing - Batch data processing
• notification_job - Scheduled notifications

Installation

Option 1: Global Installation (Recommended)

# Copy to global extensions directory
mkdir -p ~/.pi/agent/extensions/pi-dagu-workflow
cp -r src ~/.pi/agent/extensions/pi-dagu-workflow/
cp package.json ~/.pi/agent/extensions/pi-dagu-workflow/

# Or clone directly
git clone https://github.com/yourusername/pi-dagu-workflow ~/.pi/agent/extensions/pi-dagu-workflow

Then reload Pi:

/reload

Option 2: Quick Test

pi -e ./src/index.ts

Option 3: npm Package (Future)

npm install -g pi-dagu-workflow
# Add to Pi settings.json

Usage Examples

Create a Workflow

Create a Dagu workflow that:
1. Downloads data from https://api.example.com/data every hour
2. Processes it with Python
3. Uploads results to S3
4. Sends Slack notification on completion

Pi will use the dagu_create_workflow tool to generate the YAML.

Use a Template

Create an ETL pipeline workflow named "daily-sales-etl" using the etl_pipeline template.
Database host: db.company.com
Database name: sales_prod

Deploy and Run

Deploy the "daily-sales-etl" workflow and start the scheduler

Monitor

Check status of "daily-sales-etl" workflow

Requirements

• Dagu 2.0+ installed (brew install dagu-org/dagu/dagu or see https://dagu.dev)
• Pi 1.0+

File Structure

pi-dagu-workflow/
├── src/
│   └── index.ts          # Main extension file
├── package.json          # Extension metadata
└── README.md            # This file

Extension API

The extension provides these tools to Pi:

dagu_create_workflow

{
  requirements: string,      // Natural language description
  name: string,              // Workflow name
  schedule?: string,        // Cron schedule
  steps: Step[],             // Step definitions
  validate?: boolean,       // Validate after creation
  deploy?: boolean,         // Deploy to ~/.config/dagu/dags/
  env_vars?: Record<string, string>,
  notifications?: {
    on_success?: boolean,
    on_failure?: boolean,
    slack_webhook?: string,
    email?: string,
  }
}

dagu_template_workflow

{
  template: "etl_pipeline" | "ci_cd" | "backup_job" | 
            "api_monitor" | "data_processing" | "notification_job",
  name: string,
  customize?: Record<string, string>,
  deploy?: boolean
}

dagu_run_workflow

{
  workflow_name: string,
  dry_run?: boolean,
  params?: Record<string, string>
}

Configuration

The extension automatically detects Dagu installation and version on startup. It will:

• Notify if Dagu is not installed
• Show Dagu version in status bar
• Track scheduler state

Development

Testing

# Test extension with Pi
pi -e ./src/index.ts

# Or copy to extensions and reload
mkdir -p ~/.pi/agent/extensions/pi-dagu-workflow
cp src/index.ts ~/.pi/agent/extensions/pi-dagu-workflow/
cp package.json ~/.pi/agent/extensions/pi-dagu-workflow/
pi
/reload

Build (if needed)

The extension uses jiti for TypeScript, so no build step is required.

Troubleshooting

Extension not loading

• Check file location: ~/.pi/agent/extensions/pi-dagu-workflow/src/index.ts
• Run /reload in Pi
• Check Pi logs for errors

Dagu not found

• Run /dagu-install command
• Or install manually: brew install dagu-org/dagu/dagu

Scheduler won't start

• Check if port 8090 is in use: lsof -i :8090
• Check logs: cat ~/.dagu/scheduler.log

License

MIT

Contributing

PRs welcome! Please follow the Pi extension patterns.

See Also

• Dagu Documentation
• Pi Extensions Documentation
• Dagu Examples