mcp -datadog

This MCP Datadog server exposes a suite of tools that wrap Datadog APIs for use by AI assistants and code editors via the MCP interface. The available tools cover Metrics, Logs, Events, Monitors, APM/Traces, and Services dependencies, enabling you to query metrics, search logs, inspect monitors, and more through simple tool calls. For example, you can prompt your MCP client to run query_metrics with a time range to fetch CPU usage, or use search_logs with filters like service and env to retrieve log entries. The server is designed to be launched via npx (no clone or install required) or run from source if you’re developing locally, and it expects Datadog credentials to be supplied via environment variables (DATADOG_API_KEY, DATADOG_APP_KEY, and optionally DATADOG_SITE). Tools are invoked through standard MCP prompts such as datadog.query_metrics, datadog.search_logs, datadog.list_monitors, etc., depending on the capabilities you need. When using the stdio setup, ensure your client restarts after setting environment variables so the tools surface correctly.

Prerequisites:\n- Node.js 22+ (install from nodejs.org)\n- npm (comes with Node.js)\n\nOption A: Run with npx (no clone or install)\n1. Ensure you have network access.\n2. Run through your MCP client with stdio using the provided config (see mcp_config example).\n\nOption B: Run from source (development or fixed install)\n1. Clone the repository:\nbash\ngit clone https://github.com/micaelmalta/mcp-server-datadog.git\ncd mcp-server-datadog\n 2. Install dependencies:\nbash\nnpm install\n 3. Copy the example env file and populate keys:\nbash\ncp .env.example .env\n 4. Start the server:\nbash\nnpm start\n \nIn your MCP config, reference the server as a stdio type with node pointing to the entry point if you’re running from source:\njson\n{ "datadog": { "type": "stdio", "command": "node", "args": ["/path/to/mcp-server-datadog/src/index.js"], "env": { "DATADOG_API_KEY": "your_api_key", "DATADOG_APP_KEY": "your_app_key", "DATADOG_SITE": "datadoghq.com" } } }\n

Tips and troubleshooting:\n- Ensure DATADOG_API_KEY, DATADOG_APP_KEY, and optionally DATADOG_SITE are set before starting the client so tools initialize properly. Restart the MCP client after updating env vars to surface the Datadog tools.\n- The server logs tool calls to stderr as JSON lines; you can adjust MCP_SLOW_TOOL_MS to flag long-running tool calls.\n- The server does not implement rate limiting; be mindful of Datadog API quotas when performing large queries or frequent polling.\n- If a tool appears missing, double-check the MCP config references the correct tool names (e.g., query_metrics, search_logs, list_monitors, etc.) and that the environment credentials point to the intended Datadog org/site.\n- For E2E testing against Datadog, ensure your API and APP keys have sufficient permissions and the appropriate site is used (prod vs. us, eu, etc.).