Dynatrace MCP Playground

This repo is used primarily for experimenting with the Dynatrace MCP server and will be used for the DevHUB Sessions at Dynatrace Perform 2026

Follow these steps and you will learn: 👉 What is an MCP 👉 How to setup the Dynatrace MCP Server 👉 How to connect to a Dynatrace environment 👉 How to execute pre-defined prompts 👉 how to ask any question to Dynatrace

Learn more about MCPs

Check out Adriana’s blog post on to learn more about MCPs together here.

You can also check out other MCP servers here.

Why use the Dynatrace MCP Server?

This approach makes Dynatrace data accessible to team members who may not be familiar with DQL (Dynatrace Query Language) or APIs, democratizing observability data across your organization.

Key Takeaways:

• Natural Language Queries: Interact with Dynatrace without writing complex API calls or DQL queries
• Reusable Prompts: Save time with pre-built queries stored as files
• AI-Powered Analysis: Let Copilot handle complex, multi-step queries automatically

Setup

You can run this tutorial in two ways

1. Fork this GitHub repo and launch the Codespace. Once its up you are in Visual Studio and you are ready to go
2. Clone this GitHub repo locally in open it in your local Visual Studio Code instance! You will need a locally installed npx to run the Dynatrace MCP Server!

Step 1: Enable MCP support in VSCode

This tutorial uses the Visual Studio Code (VSCode) built-in GitHub Copilot chat support. Copilot supports multiple large language models (LLMs), including GPT-4 and Claude, so feel free to select your favorite.

Before you can use GitHub Copilot chat, however, you’ll need to enable MCP support in VSCode. You can do this by adding “chat.mcp.enabled”: true to your VSCode settings.json file.

Step 2: Configure Dynatrace MCP server

The LLM used by GitHub Copilot chat talks to Dynatrace via the Dynatrace MCP server. For that to happen, we must make Copilot aware of the Dynatrace MCP server, which I’ll show you below.

If you open up .vscode/mcp.json, you’ll notice the existing configuration for the Dynatrace MCP server:

{ 
  "servers": {
    // https://github.com/dynatrace-oss/dynatrace-mcp
    "dynatrace-oss/dynatrace-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
				"@dynatrace-oss/[email protected]"
			],      
      "env": {
				"DT_ENVIRONMENT": "${input:dt_environment}"
			},			
			"version": "1.0.0"