MCP ServerOfficialHTTP

Datadog MCP Server

Datadog ships an official remote MCP server that exposes observability data as tools — logs, metrics, traces, monitors and incidents. It is the difference between asking "why did latency spike at 14:05?" and clicking through six dashboards to find out.

Hosted URL

https://mcp.datadoghq.com/api/unstable/mcp-server/mcp

Suggested model

Claude Sonnet 4.5

Chat with 60+ AI models on the same workflow — switch to a different model mid-conversation and re-run the same prompt, or use Compare mode to put several side-by-side and balance quality vs. cost.

Auth

Datadog API key plus an application key, sent as headers. Use the endpoint for your region — US3, US5, EU and AP have their own hostnames.

What the Datadog MCP server does

How models use it and what it is built for.

The server turns the Datadog platform into a set of MCP tools an agent can chain. Log search accepts the same query syntax you use in the Log Explorer, scoped by time window, so an agent can narrow from a service to a single request. Metrics are queryable, which lets a model pull a timeseries and reason about the shape of it rather than describe a screenshot. APM traces are reachable for latency work, and monitors and incidents are readable so an agent can answer what is currently firing and what it is attached to. The practical value is correlation: the model can pull a monitor, fetch the logs from the same window, check the deploy events around it and propose a cause in one turn. Datadog runs regional endpoints — US1, US3, US5, EU and AP — and you must connect to the one your organisation lives in.

Tools the Datadog MCP server exposes

Typical tools an AI model can call. Exact names vary by version.

  • log search — query logs with Log Explorer syntax over a time range
  • metrics query — pull a timeseries for a metric and scope it by tag
  • APM traces — fetch traces and spans for latency and error investigation
  • monitors — list monitors, read definitions and check current state
  • incidents — read incidents and their timelines
  • events — deploy and change events for correlating a regression

Connecting to Datadog

Taken from the official Datadog documentation — see Datadog MCP Server — official documentation for the full reference.

URL format

https://mcp.{region}.datadoghq.com/api/unstable/mcp-server/mcp

Examples

US1 (default)

https://mcp.datadoghq.com/api/unstable/mcp-server/mcp

EU

https://mcp.datadoghq.eu/api/unstable/mcp-server/mcp

US3

https://mcp.us3.datadoghq.com/api/unstable/mcp-server/mcp

US5

https://mcp.us5.datadoghq.com/api/unstable/mcp-server/mcp

AP1

https://mcp.ap1.datadoghq.com/api/unstable/mcp-server/mcp

Example prompts to try

Copy any of these into MCP Agent Studio after connecting.

  • Which service accounted for the p99 latency increase between 14:00 and 14:30?

  • Show me the error logs for checkout-api in the last hour, grouped by error type.

  • What monitors are alerting right now, and were there deploys just before they fired?

  • Compare request volume for the payments service this week against last week.

Models on MCP Playground

This is not a single-model product: you get the same MCP connection with 60+ models (Claude, GPT, Gemini, DeepSeek, open-weight, and more), you can switch mid-conversation, and you can open Compare mode to run the same prompt against multiple models at once. The card above is a suggested starting point for this server — not the only choice.

Default pick for Datadog

Claude Sonnet 4.5

Incident work is multi-step correlation across logs, metrics and deploys. Sonnet 4.5 holds the thread across several tool calls where cheaper models stop at the first result.

Check an AI agent can actually use the Datadog MCP server

Listing tools proves the server is reachable, not that a model can work with it. Evals go further: they read every tool on the server, write a test suite from its real schemas, and run it — code decides pass/fail on the responses (schema conformance, error codes, pagination, result caps) while a scoring model grades plain-English tasks driven through the tools.

Get a pass/fail report per tool with the evidence behind each verdict — and replay the same suite after every schema change. Destructive tools are excluded from the run.

Run evals

Try the Datadog MCP server in your browser

Open MCP Agent Studio with the connection pre-filled. Add your token, pick any of 60+ models, and start chatting — no install required.

Open Agent Studio

Datadog MCP server — FAQ

Common questions about connecting, scoping and using it safely.

What is the Datadog MCP server?

It is Datadog’s official remote MCP server. It exposes logs, metrics, APM traces, monitors, incidents and events as MCP tools, so an AI assistant can investigate your observability data directly instead of you translating a question into a dashboard query.

Which endpoint should I use?

The one matching your Datadog site. US1 is mcp.datadoghq.com, EU is mcp.datadoghq.eu, and US3, US5 and AP1 each have their own hostname. Connecting to the wrong region fails authentication even with a valid key, which is the most common setup mistake.

What credentials does it need?

A Datadog API key and an application key. The application key carries the permissions, so scope it deliberately — an agent inherits exactly what that key can do, and read-only is the right default for investigation work.

Is the endpoint stable?

The path still contains "unstable", which is Datadog’s own signal that the API surface may change. It works today and is documented, but pin your expectations accordingly and re-check after Datadog platform updates.

Can I test it without wiring up a client?

Yes. Paste the regional URL into MCP Playground with your keys and the tools list loads in the browser. You can run a log search and see the raw JSON before deciding whether to put it in front of an agent.

Other MCP servers

More on MCP Playground

Datadog MCP Server — AI for Metrics, Logs and Incidents