Datadog MCP Server

Community

MCP server for the Datadog API, enabling incident management and more.

How to install ↓ Compare GitHub source repository ↗

Category: Dev Tools #401 of 438
Stars: ★ 143 Popular
Transport: stdio (local process)
Runtime: Node.js
Credentials: API key / credential required
License: Apache-2.0
Last commit: 4 months ago
Tools: 20

33FMRS · D

Community-maintained Datadog MCP server with broad coverage of Datadog features including incidents, monitoring, logs, and RUM, designed to be extensible. Requires proper configuration of Datadog API credentials.

Strongest · Documentation 9/20 Weakest · Reliability 4/20

Reliability

4/20

Security and permissions

6/20

Maintenance

6/20

Documentation

9/20

Setup experience

8/20

Why each score

Reliability 4/20: Only the README and repository metadata were supplied; no source, package manifest, CI workflow, or tests are available. The README tool list is untrusted evidence, so the init handshake, actual tool availability, and behavioral consistency cannot be verified. Per static calibration, missing executable evidence caps the score; this is a deduction for missing evidence, not proven breakage.
Security and permissions 6/20: The README indicates API/App keys are injected via environment variables, which is positive; however, no source is available to audit for extra network calls, secret leakage, or privilege overreach. State-changing tools such as mute/unmute/downtime have no described confirmation flow, and least-privilege or data-boundary guidance is absent. No red-line issue was found, but security evidence is insufficient, so a deduction applies.
Maintenance 6/20: The repository is not archived and declares an Apache-2.0 license, which are basic maintenance signals; however, no commit history, releases, dependency updates, or security-response channel are in evidence, and 28 open issues remain. CI/coverage badges in the README are untrusted and do not count as executable evidence, so maintenance is scored moderately low.
Documentation 9/20: The README covers authentication, installation, parameters and returns for 20 tools, and debugging, with good structure; however, it contains a clear copy-paste error: a GitHub server configuration block appears in the Claude Desktop section before the Datadog example, and cost, limits, troubleshooting, and usage limitations are missing. The README itself is untrusted evidence, so accuracy cannot be validated.
Setup experience 8/20: The docs provide pnpm manual build, npx and Smithery install paths, and Claude Desktop config examples, which are clear; however, no package.json, build artifacts, or CI evidence verifies these steps, the GitHub config block may mislead users, and the environment-variable setup is non-trivial. The static cap is 15, and uncertainty further lowers the score.

Static review · not runListed 2026-08-07

Read the FMRS scoring method →

Fit and risk

What it can accessUses the networkChanges third-party account data

Best for

Developers and SREs looking to integrate Datadog operations into AI-assisted workflows
Teams using Claude Desktop and other MCP-compatible clients

Not for

Users seeking official Datadog support
Those needing advanced Datadog API operations beyond the current tool set
Users unfamiliar with Datadog API authentication

Required permissions

Requires Datadog API key and application key with permissions for the used APIs, such as incidents, monitors, logs, dashboards, metrics, APM, hosts, downtimes, and RUM.

Risks and side effects

Community-maintained, unofficial, may have bugs or unexpected behavior
MCP server runs with persistent API keys, which can be leaked if not secured
Some operations like muting hosts or scheduling downtimes can have significant impact if misconfigured
API calls may consume Datadog quota or incur costs

Setup

Before you start

Runtime：Node.js

DATADOG_API_KEY requiredsecret Datadog API key used to call the Datadog API; obtain it from your Datadog organization's API Keys page.

DATADOG_APP_KEY requiredsecret Datadog Application key needed for app-level resources like incidents and dashboards; obtain it from your Datadog organization's Application Keys page.

Other optional settings (3)

DATADOG_SITE optional Optional Datadog site (e.g. datadoghq.eu); defaults to datadoghq.com.

DATADOG_SUBDOMAIN optional Optional Datadog subdomain (e.g. <your-subdomain>.datadoghq.com) used for generated links.

DATADOG_STORAGE_TIER optional Optional logs storage tier for v2 log searches; one of indexes, online-archives, or flex.

Ensure Node.js and pnpm are installed.
Clone the repository and run pnpm install and pnpm build.
Set environment variables DATADOG_API_KEY and DATADOG_APP_KEY (required), and optionally DATADOG_SITE, DATADOG_SUBDOMAIN, and DATADOG_STORAGE_TIER.
Configure the server in your MCP client using npx -y @winor30/mcp-server-datadog or point to the built build/index.js.
For automated installation, run npx -y @smithery/cli install @winor30/mcp-server-datadog --client claude.

claude_desktop_config.json

{
  "mcpServers": {
    "mcp-server-datadog": {
      "command": "npx",
      "args": [
        "-y",
        "@winor30/mcp-server-datadog"
      ],
      "env": {
        "DATADOG_API_KEY": "<YOUR_API_KEY>",
        "DATADOG_APP_KEY": "<YOUR_APP_KEY>",
        "DATADOG_SITE": "<YOUR_SITE>",
        "DATADOG_SUBDOMAIN": "<YOUR_SUBDOMAIN>",
        "DATADOG_STORAGE_TIER": "indexes"
      }
    }
  }
}

Shown for Claude Desktop. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.

.vscode/mcp.json

{
  "servers": {
    "mcp-server-datadog": {
      "command": "npx",
      "args": [
        "-y",
        "@winor30/mcp-server-datadog"
      ],
      "env": {
        "DATADOG_API_KEY": "<YOUR_API_KEY>",
        "DATADOG_APP_KEY": "<YOUR_APP_KEY>",
        "DATADOG_SITE": "<YOUR_SITE>",
        "DATADOG_SUBDOMAIN": "<YOUR_SUBDOMAIN>",
        "DATADOG_STORAGE_TIER": "indexes"
      }
    }
  }
}

Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).

Terminal

claude mcp add mcp-server-datadog -e 'DATADOG_API_KEY=<YOUR_API_KEY>' -e 'DATADOG_APP_KEY=<YOUR_APP_KEY>' -e 'DATADOG_SITE=<YOUR_SITE>' -e 'DATADOG_SUBDOMAIN=<YOUR_SUBDOMAIN>' -e DATADOG_STORAGE_TIER=indexes -- npx -y @winor30/mcp-server-datadog

Run it in a terminal; replace any <…> placeholders with your own values first.

Generate and validate client config

Check that it works

Check that the client's tool list includes tools like get_monitors or list_incidents from mcp-server-datadog, then ask for the active hosts count — a valid number confirms the connection works.

Troubleshooting

Verify DATADOG_API_KEY and DATADOG_APP_KEY are set correctly and have sufficient permissions.
Check DATADOG_SITE and DATADOG_SUBDOMAIN match your Datadog account.
Inspect MCP server logs, and use MCP Inspector (`npm run inspector`) for detailed debugging.
Ensure the server executable path is correct or that `npx` commands can access the network.

Things to try

Once connected, you can ask your AI assistant things like:

List my Datadog monitors currently in alert state.
Search Datadog logs for 'error' over the last 24 hours.
Schedule a downtime for host:my-host starting in 2 hours today.
Show the active hosts count and list hosts active in the last 2 hours.

Tools 20

list_incidents read-only

Retrieve a list of incidents from Datadog with optional filters and pagination.

get_incident read-only

Retrieve detailed information about a specific Datadog incident.

get_monitors read-only

Fetch the status of Datadog monitors, filterable by state, name, and tags.

get_logs read-only

Search and retrieve logs from Datadog using a query and time range.

list_dashboards read-only

Get a list of dashboards from Datadog, filterable by name and tags.

get_dashboard read-only

Retrieve a specific dashboard from Datadog by ID.

query_metrics read-only

Retrieve metrics data from Datadog for a specified time range.

list_traces read-only

Get a list of APM traces with query, time range, service, operation, and sorting filters.

Show 12 more tools

list_hosts read-only

Get a list of hosts from Datadog with filtering, sorting, pagination, and metadata options.

get_active_hosts_count read-only

Get the total number of active hosts in Datadog.

mute_host writes

Mute a host in Datadog with optional message and end time.

unmute_host writes

Unmute a host in Datadog.

list_downtimes read-only

List scheduled downtimes from Datadog, optionally filtering by current only or monitor ID.

schedule_downtime writes

Schedule a downtime in Datadog with scope, start/end times, message, timezone, monitor, and recurrence settings.

cancel_downtime destructive

Cancel a scheduled downtime in Datadog by ID.

get_rum_applications read-only

Get all RUM applications in the organization.

get_rum_events read-only

Search and retrieve RUM events from Datadog using a query and time range.

get_rum_grouped_event_count read-only

Search, group, and count RUM events by a specified dimension.

get_rum_page_performance read-only

Get page (view) performance metrics from RUM data, including average, min, max, and count.

get_rum_page_waterfall read-only

Retrieve RUM page (view) waterfall data filtered by application name and session ID.

Use cases

Manage Datadog incidents directly within your MCP client

Quickly query Datadog monitors, logs, dashboards, and metrics

Perform host mute/unmute and downtime scheduling without switching context

Monitor user experience through RUM tools

Supported clients

Claude Desktop

Listed from the project's documentation, not tested by this site.

Overview

mcp-server-datadog is a community-maintained MCP server that integrates Datadog's observability capabilities into the MCP ecosystem. It provides a set of tools for managing incidents, retrieving monitors, logs, dashboards, metrics, APM traces, and managing hosts and downtimes, as well as RUM (Real User Monitoring) features. This server is developed independently and is not officially endorsed by Datadog.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →

NPM Sentinel MCP Server 79 · B

AI-powered NPM package analysis MCP server

★ 18 · Tools 19 Compare with this →

MartinLoop MCP Server 79 · B

An independent control layer that gives AI coding agents budgets, brakes, and receipts

★ 413 Compare with this →

GitHub MCP Server 76 · B

GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language

★ 33.7k · Tools 15 Compare with this →

Compare with other MCP servers

Source revision 24f2c5472861 Data synced 2026-10-19 Read the FMRS scoring method