@dynatrace-oss/dynatrace-mcp-server

What it does

The server bridges AI assistants and Dynatrace observability data. It exposes problem management, data querying (DQL), entity discovery, and automation capabilities through MCP tools. Claude and other AI clients can fetch production metrics, logs, exceptions, and real-time observability signals directly in conversation.

Who it's for

DevOps engineers, SREs, and backend engineers who want to ask Claude about production issues, incidents, or deployment health using real Dynatrace data—without leaving their editor or IDE.

Common use cases

• Query logs and traces with natural language; ask Claude to "show error logs from the past hour" and have it generate and execute DQL statements
• Investigate incidents by fetching problems, exceptions, and related spans for root-cause analysis with full production context
• Check deployment health metrics before or after releases to gate promotion or trigger rollback
• Analyze vulnerabilities and get AI-powered breach impact assessment from Dynatrace security data
• Integrate Slack or email notifications so critical issues detected in Dynatrace surface immediately in your workflow

Setup pitfalls

• Node.js v22.10 or newer is required; older versions will not run the server
• DT_ENVIRONMENT must be set to the Dynatrace Platform URL (e.g., https://abc12345.apps.dynatrace.com), not the deprecated classic format
• OAuth login initiates in your browser on startup; ensure the browser can reach your Dynatrace environment or authentication will hang
• The repository is in maintenance mode with a failing CI pipeline; verify the latest npm package before deploying to production