Install guide and config at curatedmcp.com

Datadog MCP: Let Claude Debug Your Stack in Real Time

If you're tired of tab-switching between your code editor, Datadog dashboards, and logs trying to figure out what broke, there's a better way. The official Datadog MCP server connects Claude (or your favorite AI agent) directly to your full observability stack—metrics, logs, traces, and incidents—turning "something is wrong" into "here's why and how to fix it" within the same conversation.

What It Does

Datadog MCP is a Model Context Protocol server that gives AI agents read and write access to your entire observability layer. Query metrics with PromQL/DQL expressions, tail logs in real time, inspect distributed traces and service maps, check monitor status, and pull incident timelines—all programmatically.

The magic is in the workflow: when an alert fires, Claude can autonomously fetch the metric spike, pull correlated logs from the same time window, trace the request across services, and propose a root cause and fix—without you context-switching. You get structured data from your observability platform piped directly into your AI's reasoning loop.

Features include APM service performance queries, infrastructure and host health inspection, synthetic test result analysis, and the ability to create dashboards and widgets on the fly. For teams already invested in Datadog, this eliminates the friction of manual investigation.

How to Install

Install the server via npm:

npx -y @datadog/mcp-server

Then add it to your Claude Desktop configuration:

{
  "mcpServers": {
    "datadog-mcp": {
      "command": "npx -y @datadog/mcp-server",
      "env": {
        "DATADOG_API_KEY": "your_api_key_here",
        "DATADOG_APP_KEY": "your_app_key_here",
        "DATADOG_SITE": "datadoghq.com"
      }
    }
  }
}

Grab your API and app keys from your Datadog account settings. Full setup guides for Cursor and Windsurf are in the docs.

Real-World Use Cases

• Incident triage in Claude: Paste an alert into your chat. Claude queries the metric, pulls logs, finds the trace, and hands you a diagnosis and diff—all in one response.
• On-call automation in Cursor: Your CI/CD alert hits. Ask Claude to check host health, check related monitors, and suggest scaling actions—all pulling live data from your infrastructure map.
• Postmortem data gathering: Generate a timeline of events, correlated logs, and affected services from an incident—Claude assembles the narrative without you hunting through tabs.

Full install guides for Claude Desktop, Cursor, Windsurf, and more at CuratedMCP.