Docker's New MCP Toolkit and Catalog Explained in 5 Minutes

Docker has recently announced two new powerful tools for developers: the MCP Catalog and the MCP Toolkit. These additions aim to streamline the use of Docker images that are compatible with the MCP (Mergify Command Palette) standard.

Introducing the MCP Catalog

The MCP Catalog is a centralized collection of all Docker images that are MCP-compatible. Essentially, Docker has gathered all the MCP servers that can be pulled from their registry into one convenient location. This catalog includes a wide variety of tools, such as the mcp-playwright server, which we will explore in a practical demonstration later in this article.

Currently, the catalog features more than 120+ MCP servers, providing a rich ecosystem of tools for various development needs.

The MCP Toolkit

The second major announcement is the availability of the MCP Toolkit. This toolkit allows you to install and manage multiple MCP servers seamlessly. You can enable one or all of them to access a wide range of tools directly within your development environment.

A key feature of the toolkit is its configuration management. You can securely store credentials and tokens for different services. For instance, if you are using a tool that requires GitHub credentials, you can configure them within the toolkit to run in a secured environment.

Installation Note: To install the MCP Toolkit, you can navigate to the extensions marketplace in your IDE (like Visual Studio Code) and search for "MCP toolkit." Once installed, you can easily enable or disable the tools you need.

Practical Demo: Running a Playwright Server

Let's walk through a quick demonstration of how to run an MCP server. For this example, we'll use the Playwright MCP server, which provides access to 25 different tools.

Step-by-Step Guide:

1. Open your project in Visual Studio Code.
2. Press Ctrl+Shift+P to open the command palette.
3. Type and select the add server command.
4. Run the following command to start the Playwright server within Docker: bash docker run mcp-playwright
5. Assign a name to the server instance and save it in your current workspace.

The server will start running immediately. Your settings should reflect that mcp-playwright is active with the name you provided.

Interacting with the Server:

Once the server is running, you can switch to the chat mode in your IDE to interact with it. For example, you can issue a command to Playwright.

Let's try the following prompt:

"Go to bing.com and find the latest news about AI"

The system will then perform the following actions: - Action 1: Confirm that you want to navigate to bing.com. You can approve this. - Action 2: A browser window will open. - Action 3: The tool will ask for permission to search the web. After you approve, it will execute the search for "AI latest news."

The process will finish, and you will have successfully used the new Docker toolkit with an MCP-compatible server to automate a browser task.