Datadog

This MCP (Model Context Protocol) server lets you interact with Datadog APIs through LLMs. It provides a robust, production-ready integration with 351 tools that support CRUD operations, curated capabilities, and automatically generated endpoints from the Datadog schema. You can manage dashboards, monitors, downtimes, users, teams, and more, while enjoying strong validation, schema autodiscovery, progress tracking, and a feature-rich CLI for debugging and management.

How to use

You use this MCP server by running it locally or connecting to it via an MCP-compatible client. The server exposes a large suite of tools that let your LLMs perform actions against Datadog, including creating, reading, updating, and deleting core resources, as well as running advanced queries and maintenance tasks. To start using it, launch the MCP server in your environment and configure your MCP client to point to the running process. The client will discover available tools, fetch schemas, validate inputs, and automatically construct valid API calls based on the tool schemas.

Key capabilities you will leverage include autodescovery of schema, robust input validation, real-time progress tracking for long-running operations, intelligent error handling with retries, and a rich CLI for inspection and debugging.

How to install

Prerequisites you need before running the server are listed here. You will work with Node.js and Datadog API credentials.

# Install the MCP Datadog server globally
npm install -g mcp-datadog-server

# Or install locally by cloning the repository and installing dependencies
git clone https://github.com/ClaudioLazaro/mcp-datadog-server.git
cd mcp-datadog-server
npm install

Optionally, you can start a local development server using the standard commands shown below.

# Start the MCP server (standard)
mcp-datadog-server serve

# Alternative start via npm script
npm start

# With filters for specific folders
mcp-datadog-server serve --folders=Dashboards,Monitors

Configuration highlights

To connect to Datadog, you must provide your Datadog API and Application keys. You can place these in environment variables or in a .env file for convenience.

# Environment variables (required)
DD_API_KEY=your_api_key
DD_APP_KEY=your_app_key

# Optional site configuration (default is datadoghq.com)
DD_SITE=datadoghq.com
DD_SUBDOMAIN=api
MCP_DD_FOLDERS=Dashboards,Logs
MCP_DD_SCHEMA_PATH=./schema.json
MCP_DD_MAX_RETRIES=3
MCP_DD_RETRY_BASE_MS=1000
MCP_DD_TIMEOUT_MS=30000
MCP_DD_USER_AGENT=mcp-datadog

Additional usage notes

You can integrate with Claude Desktop, npx, or a local Node.js command to drive the MCP server from your LLM workflow. Examples of common configurations are provided for different runtimes.

Available tools

create_monitor

Create a new monitor in Datadog with validation against the schema.

get_monitor

Fetch a monitor by its ID.

update_monitor

Update an existing monitor by ID.

delete_monitor

Delete a monitor by ID.

list_monitor

List all monitors.

create_dashboard

Create a new Datadog dashboard.

get_dashboard

Get a dashboard by ID.

update_dashboard

Update a dashboard by ID.

delete_dashboard

Delete a dashboard by ID.

list_dashboard

List all dashboards.

create_downtime

Schedule a downtime window.

get_downtime

Retrieve a scheduled downtime by ID.

update_downtime

Update a scheduled downtime.

delete_downtime

Cancel a scheduled downtime.

list_downtime

List all downtimes.

create_user

Create a new Datadog user.

get_user

Get a user by ID.

update_user

Update a user by ID.

delete_user

Delete a user by ID.

list_user

List all users.

create_team

Create a new team.

get_team

Get a team by ID.

update_team

Update a team by ID.

delete_team

Delete a team by ID.

list_team

List all teams.

list_dashboards

List dashboards with advanced filters and pagination.

search_logs

Perform advanced log searches with filters.

query_metrics

Query metrics timeseries data.

manage_incidents

Full management of incidents.

manage_synthetics

Full management of synthetic tests.