datadog MCP Server

Use This MCP server To

Retrieve Kubernetes logs from Datadog for AI analysis Feed real-time operational data into AI copilots Enhance incident response with AI-driven log insights Integrate Datadog logs into Claude desktop workflows Automate log data retrieval for troubleshooting Provide structured log context to language models

README

datadog-mcp

Model Context Protocol server that retrieves logs from a specific Kubernetes cluster and namespace from Datadog for analysis by Claude.

Prerequisites

Datadog API Key
Datadog Application Key with permission to read logs
Claude Desktop application
Node.js >= 18

Usage

Clone the repository
Build the application

cd datadog-mcp
npm run build

Add the Datadog API Key and Application Key to your Claude desktop configuration, and MCP server configuration if needed. Example for macOS:

vim  ~/Library/Application Support/Claude/claude_desktop_config.json

The above file should look like the following. In this example we are using nvm, so we need to provide the full path to the executable in line 4. If you are not using nvm, you can just pass node.

{
  "mcpServers": {
     "datadog-mcp": {
        "command": "/Users/<your-user>/.nvm/versions/node/v22.11.0/bin/node",
        "args": [
          "/<path-to-repo>/datadog-mcp/build/index.js"
        ],
        "env": {
          "DD_API_KEY": <YOUR_API_KEY>
          "DD_APP_KEY": <YOUR_APP_KEY>
        }
     }
  }
 }

Open the Claude desktop application. On the bottom right hand corner of the chat window you should see a hammer 🔨, and 1 MCP tool available when hovering. If you don't see the hammer, something is not working properly.
Ask Claude to provide information about a specific cluster and namespace. For example:

What insights can you give me about datadog logs in the cluster dev and namespace app?

Claude will use your API and APP keys, make a request to Datadog, parse the logs and provide a response.

To do

Pass start and end timestamp
Query Metrics
Query Monitors

Reference

The code is based on the starter example provided by Anthropic.

datadog-mcp FAQ

What are the prerequisites for running datadog-mcp?

You need a Datadog API Key, an Application Key with log read permissions, Claude Desktop, and Node.js version 18 or higher.

How do I configure datadog-mcp with Claude Desktop?

Add your Datadog API and Application Keys to the Claude desktop configuration file, specifying the MCP server command and arguments as per the README instructions.

Can datadog-mcp retrieve logs from any Kubernetes cluster?

It retrieves logs from Kubernetes clusters and namespaces configured within your Datadog account, based on the API keys and permissions provided.

Is datadog-mcp compatible with other AI models besides Claude?

While designed for Claude, datadog-mcp can be adapted to work with other MCP-compatible AI models like OpenAI's GPT-4 and Anthropic's Claude by adjusting configuration.

What Node.js version is required to run datadog-mcp?

Node.js version 18 or higher is required to build and run the datadog-mcp server.

How do I build the datadog-mcp server?

Clone the repository and run 'npm run build' inside the datadog-mcp directory to compile the application.

Where do I specify the path to the Node.js executable in the configuration?

In the Claude desktop configuration JSON, specify the full path to the Node.js executable if using version managers like nvm, or simply 'node' if not.

Does datadog-mcp support real-time log streaming?

It retrieves logs for analysis but depends on Datadog's API capabilities; real-time streaming depends on Datadog's log ingestion and API refresh rates.