home / mcp / datadog mcp server

DataDog MCP Server

This project is an early-stage Model Context Protocol (MCP) server in Go, exposing tools for LLMs to interact with Datadog’s APIs (V1/V2). It provides a generated client base but is not yet a full implementation. Next steps include extending the toolset.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "believe-sa-datadog-mcp": {
      "command": "./build/datadog-mcp-server",
      "args": [],
      "env": {
        "DD_API_KEY": "YOUR_DATADOG_API_KEY",
        "DATADOG_SITE": "datadoghq.com",
        "DATADOG_APP_KEY": "YOUR_DATADOG_APP_KEY"
      }
    }
  }
}

You can connect AI assistants to DataDog's observability suite through a dedicated MCP server. This server exposes DataDog resources like dashboards, events, metrics, and logs via a standardized interface, enabling intelligent agents to monitor, analyze, and automate across your DataDog environment.

How to use

Use an MCP client to interact with the DataDog MCP Server. You can query dashboards, retrieve host status, manage events for incident tracking, and access logs, traces, and performance metrics. You can also automate operations by interacting with monitors, downtimes, and alerts. The server is designed to provide consistent, real-time access to DataDog data through the MCP layer, so your AI assistants can work with data and actions from DataDog as if they were native tools.

Typical usage patterns include: - Asking for a list of dashboards and their current statuses. - Fetching recent events or creating a new incident during an AI-generated remediation flow. - Pulling latest logs or traces related to a given host or service. - Triggering a monitor or applying a downtime in response to detected anomalies.

How to install

Prerequisites you need before installation.

# Build the MCP server
make build

# Configure DataDog API access
export DD_API_KEY="your-datadog-api-key"
export DATADOG_APP_KEY="your-datadog-app-key"  # Optional
export DATADOG_SITE="datadoghq.eu"  # or datadoghq.com

# Generate MCP configuration
make create-mcp-config

# Run the server locally
./build/datadog-mcp-server

Configuration and runtime details

Environment variables required by the DataDog MCP Server are provided during setup. You should set these in your shell or environment manager before starting the server.

Key environment variables shown in the quickstart include: - DD_API_KEY: DataDog API key. - DATADOG_APP_KEY: DataDog application key (optional). - DATADOG_SITE: Datadog site to use (datadoghq.eu or datadoghq.com).

Notes and next steps

The server is shipped with a set of implemented MCP tools and a path to generate an MCP configuration that wires your client to these tools. You can extend functionality as DataDog tools are progressively released or as you adopt new MCP capabilities.

Available tools

v1_list_dashboards

List all dashboards available in DataDog via the MCP interface.

v1_get_dashboard

Retrieve detailed information for a specific dashboard by ID.

v1_list_events

List events in DataDog for incident tracking and analysis.

v1_create_event

Create a new event in DataDog to capture incidents or notable activity.

v1_test_connection

Test connectivity between the MCP server and the DataDog platform to verify API access.