Datadog Logs MCP Server

An MCP (Model Context Protocol) server for searching Datadog logs via HTTP.

Features

• Search Datadog logs with customizable queries

• Specify time ranges for log searches

• Control result limits

• Runs as an HTTP server on port 5000

Prerequisites

• Docker and Docker Compose

• Datadog API Key

• Datadog Application Key

Setup

1. Copy the example environment file and add your Datadog credentials:

2. Edit .env and add your actual Datadog keys:

Running with Docker Compose

Start the server:

The server will be available at http://localhost:5000

To run in detached mode:

To stop the server:

API Endpoints

SSE Connection

• GET /sse - Establish SSE connection for MCP communication

Message Endpoint

• POST /messages - Send MCP protocol messages

MCP Tool: search_logs

Search Datadog logs with the following parameters:

• query (required): Log search query (e.g., "env:prd AND service:pms-connectors")

• from (required): Start time (e.g., "now-10m", "2024-01-01T00:00:00Z")

• to (required): End time (e.g., "now", "2024-01-01T01:00:00Z")

• limit (optional): Maximum number of logs to return (default: 10)

Example Usage

The server will search Datadog logs using the API endpoint:

With the query parameters provided through the MCP tool invocation.

Development

To build locally without Docker:

Environment Variables

• DD_API_KEY: Your Datadog API key (required)

• DD_APPLICATION_KEY: Your Datadog application key (required)

• PORT: Server port (default: 5000)