Home Phase 5 GUI
Phase 5 of 6 🖱️ GUI Method

Agent Execution Verification

Verify Daytona integration and monitor agent performance through web dashboards

1 day
Beginner Level
100% GUI
🤖 What is Agent Execution?
Your Helium AI agents run in isolated Daytona sandboxes. This phase verifies the integration works correctly and sets up monitoring.

Prerequisites

Step 1: Verify Daytona Configuration

1.1

Access Daytona Dashboard

  • 1
    Go to Daytona Dashboard
  • 2
    Log in with your credentials
  • 3
    Click "Settings" or "API Keys"
  • 4
    Verify your API key is active
  • 5
    Copy the API key (if you need to update it)
1.2

Verify API Key in AWS Secrets Manager

AWS Console Navigation
AWS Console → Search "Secrets Manager" → Secrets Manager
  • 1
    Find secret: helium/backend/production
  • 2
    Click on the secret name
  • 3
    Click "Retrieve secret value"
  • 4
    Verify these keys exist:
    • DAYTONA_API_KEY
    • DAYTONA_SERVER_URL
    • DAYTONA_TARGET
  • 5
    If missing, click "Edit" and add them

Step 2: Test Agent Execution

2.1

Test from Your Application

Since agent execution happens through your application, test it directly:

  • 1
    Go to your frontend: https://he2.ai
  • 2
    Log in to your account
  • 3
    Create a new agent or use existing agent
  • 4
    Send a test message: "Create a simple Python script"
  • 5
    Verify the agent responds and executes the task
✅ Success Indicators:
• Agent responds within 5-10 seconds
• Sandbox environment is created
• Code execution completes
• Results are returned to frontend

Step 3: Set Up CloudWatch Monitoring

3.1

Create CloudWatch Dashboard

AWS Console Navigation
AWS Console → Search "CloudWatch" → CloudWatch → Dashboards
  • 1
    Click "Create dashboard"
  • 2
    Dashboard name: helium-agent-monitoring
  • 3
    Click "Create dashboard"
  • 4
    Click "Add widget"
  • 5
    Select Line widget type
  • 6
    Data source: Metrics
  • 7
    Add metrics for ECS service (CPU, Memory, Task count)
  • 8
    Click "Create widget"
3.2

Set Up CloudWatch Alarms

AWS Console Navigation
CloudWatch → Alarms → All alarms
  • 1
    Click "Create alarm"
  • 2
    Click "Select metric"
  • 3
    Browse: ECS → ClusterName, ServiceName
  • 4
    Select CPUUtilization metric
  • 5
    Click "Select metric"
  • 6
    Conditions:
    • Threshold type: Static
    • Whenever CPUUtilization is: Greater than
    • Than: 80
  • 7
    Click "Next"
  • 8
    Configure SNS notification (optional)
  • 9
    Alarm name: helium-backend-high-cpu
  • 10
    Click "Create alarm"
Recommended Alarms
Create alarms for:
• High CPU (>80%)
• High Memory (>80%)
• Task count (< desired count)
• API errors (>5% error rate)

Step 3: Monitor Agent Logs

3.1

Access CloudWatch Logs

AWS Console Navigation
CloudWatch → Logs → Log groups
  • 1
    Find log group: /ecs/helium-backend
  • 2
    Click on the log group
  • 3
    Click on any log stream to view logs
  • 4
    Look for agent execution logs
    Search for keywords like "agent", "daytona", "sandbox"
Using CloudWatch Logs Insights
Click "Logs Insights" in the left sidebar to run queries:
fields @timestamp, @message
| filter @message like /agent/
| sort @timestamp desc
| limit 20
3.2

Monitor in Daytona Dashboard

  • 1
    Go back to Daytona Dashboard
  • 2
    Click "Workspaces" or "Sandboxes"
  • 3
    View active sandboxes created by your agents
  • 4
    Check sandbox status and resource usage
  • 5
    Review execution logs in Daytona

Phase 5 Verification Checklist

✅ Phase 5 Complete! Agent execution is verified and monitored. Ready for Phase 6!
Back to Phase 4 Next: Phase 6 - CI/CD Pipeline