datadog

This MCP server provides a Datadog tracing and log analysis workflow that lets AI assistants query Datadog APM for error traces, search logs, and generate diagnostic workflows. You can invoke tools to list error traces for a service, inspect a specific trace in detail (including spans, logs, and a generated debugging workflow), extract structured test scenarios from traces, and correlate logs with traces to build a holistic debugging picture. The server is designed to reduce token usage by performing heavy lifting server-side, returning structured results that an AI assistant can reason over. To use it, run the Java-based MCP server (via the provided jar) and configure Claude Code’s mcp.json to point to the running instance. Once operational, you can call tools like trace.list_error_traces, trace.inspect_error_trace, trace.extract_scenario, log.search_logs, and log.correlate to interact with Datadog data programmatically.

Prerequisites:

• Java 21 or newer installed on the host
• Access to Datadog (optional API keys for certain features)
• Internet access to fetch dependencies (if running the installer or pulling the JAR)

Installation options:

Option A: Run the JAR directly

1. Ensure you have Java 21+ installed

2. Obtain the datadog-mcp-server.jar (from your build artifact or release page)

3. Start the MCP server:

   java -jar datadog-mcp-server.jar

4. If needed, set environment variables (for example, DATADOG_API_KEY) before starting the process

Option B: Use the provided installer script (as described in README)

1. Prerequisites: curl, bash, and Java 21+ installed

2. Run the installer which builds and copies the JAR to the Claude environment:

   curl -fsSL https://raw.githubusercontent.com/waabox/datadog-mcp-server/main/install.sh | bash

3. Follow prompts to provide your Datadog API keys (optional) and complete installation

4. Configure Claude Code's mcp.json to reference the running MCP server

Note: The server is Java-based and expects a JAR artifact. If you deploy in a container or cloud environment, ensure port exposure and proper environment variables are configured.

Tips and caveats:

• Datadog API access: Some features may require an active Datadog API key. If you don’t provide one, optional features may be disabled.
• Environment labeling: Use the env parameter (prod, staging, dev) consistently when querying traces to keep results aligned with your Datadog data.
• Performance: Server-side processing reduces token usage for AI reasoning but depends on the Datadog API response times. For large time windows, consider narrowing the range with from/to and limit parameters.
• Troubleshooting: If the MCP server fails to start, check Java version compatibility, ensure the JAR path is correct, and verify network access to Datadog endpoints. Logs will help identify missing environment variables or authentication issues.