Agent Operations Commander

• 

  Real Dynatrace MCP investigation mode connected to live service telemetry.

• 

  Solution architecture showing Google ADK, Dynatrace MCP, and the optional incident response console.

• 

  Public GitHub repository containing the complete Agent Operations Commander source code.

• 

  Google Agent Development Kit used to build and orchestrate the AI operations agent.

• 

  AI-generated incident analysis with root cause, evidence, recommendations, and timeline.

• 

  Dynatrace integration discovered through Google Cloud Marketplace resources.

• 

  Dynatrace resources and credentials configured within the Google Cloud project environment.

Inspiration

Operations teams often spend too much time switching between dashboards, logs, metrics, and alerts before they can identify the cause of an incident.

I wanted to see if an AI agent could act as an operations teammate, helping engineers investigate incidents, understand telemetry, identify root causes, and recommend actions.

That idea became Agent Operations Commander.

What it does

Agent Operations Commander is an AI-powered incident response agent built with Google ADK, Gemini, and Dynatrace MCP.

The agent investigates incidents, analyzes telemetry, identifies likely root causes, presents evidence, recommends actions, and generates incident timelines.

How I built it

• Google ADK for agent orchestration
• Gemini for reasoning and analysis
• Dynatrace MCP for observability context
• React, Vite, and CopilotKit for the operator console
• FastAPI for backend services

The architecture connects a Google ADK agent to Dynatrace MCP tools through a React-based operations dashboard.

Challenges I ran into

The biggest challenge was integrating multiple technologies into a single workflow while keeping the user experience simple.

I also spent considerable time refining the dashboard, investigation workflows, and MCP integration.

Accomplishments that I'm proud of

• Built a Google ADK incident-response agent
• Integrated Dynatrace MCP workflows
• Created demo and real investigation modes
• Built a React and CopilotKit operations console
• Generated structured incident reports with evidence and remediation actions

What I learned

I learned how to build agentic workflows using Google ADK and how MCP can connect AI agents to external operational tools.

I also learned the importance of presenting investigation results in a format that operators can quickly understand and act on.

What's next for Agent Operations Commander

Future work includes additional MCP integrations, multi-agent investigations, production deployment, and expanded operational tooling support.

I would also like to extend the platform with additional observability, cloud operations, and incident-management integrations through MCP, allowing the agent to investigate incidents across a broader operational environment.

Built With

• adk
• ag-ui
• apis
• copilotkit
• dynatrace
• fastapi
• gemini
• github
• javascript
• mcp
• ngrok
• python
• react
• rest
• vite