Mcp Datadog Server

What is Mcp Datadog Server?

Mcp Datadog Server is a production‑ready MCP (Model Context Protocol) server that provides 351 tools for interacting with Datadog APIs through LLMs. It includes curated tools, CRUD operations for core resources, and automatically generated tools from the official Datadog schema. It is designed for developers and SRE teams who want to manage Datadog resources using natural language.

How to use Mcp Datadog Server?

Install globally via npm (npm install -g mcp-datadog-server) or clone the repository and run npm install. Set the required environment variables DD_API_KEY and DD_APP_KEY (plus optional DD_SITE, etc.). Start the server with mcp-datadog-server serve or integrate it with LLMs like Claude Desktop by adding an MCP server entry in the configuration file.

Key features of Mcp Datadog Server

• 351 tools total: 9 curated, 25 CRUD, 319 generated from schema
• Automatic parameter discovery by LLMs via JSON Schema
• Robust validation with Zod schemas
• Progress tracking and intelligent error handling with retry
• Rich CLI for tool listing, inspection, and validation

Use cases of Mcp Datadog Server

• Create, update, and delete Datadog monitors using natural language
• Manage dashboards, downtimes, users, and teams via LLM conversations
• Search logs and query metrics with complex filters
• Handle incidents and synthetic tests through automated tool calls
• Filter tool categories to focus on specific Datadog resources (e.g., Dashboards, Logs)

FAQ from Mcp Datadog Server

How does the LLM discover tool parameters?

The LLM automatically lists all available tools, fetches each tool’s full JSON Schema, and constructs valid calls with the correct parameter names, types, and constraints – without needing prior knowledge of the parameters.

What environment variables are required and optional?

Required: DD_API_KEY (Datadog API key) and DD_APP_KEY (Datadog application key). Optional: DD_SITE (default datadoghq.com), DD_SUBDOMAIN, MCP_DD_FOLDERS (filter categories), MCP_DD_SCHEMA_PATH, MCP_DD_MAX_RETRIES, MCP_DD_RETRY_BASE_MS, MCP_DD_TIMEOUT_MS, and MCP_DD_USER_AGENT.

How can I filter which tools are loaded?

Set the environment variable MCP_DD_FOLDERS to a comma‑separated list of categories (e.g., Dashboards,Monitors,Logs) to load only tools from those categories. You can also use the --folders flag when starting the server.

What rate limiting handling is included?

The server automatically respects Datadog’s retry-after headers and uses configurable retry logic (MCP_DD_MAX_RETRIES and MCP_DD_RETRY_BASE_MS) to handle rate limits gracefully.

How do I debug “Tool not found” or “Invalid API key” errors?

Use mcp-datadog-server list-tools --detailed | grep <tool_name> to check if the tool exists. Validate credentials with mcp-datadog-server validate. Ensure DD_API_KEY and DD_APP_KEY are set correctly in your environment.