MCP Fundamentals

In this video, you will set up the MCP Inspector. This is a developer tool that allows you to test, debug, and interact with your MCP servers without needing to run a full AI client like Claude Desktop.

The MCP Inspector is a web-based tool that runs via npx, which comes bundled with Node.js. So, the first step is to ensure you have Node.js installed.

Open your browser and navigate to nodejs.org.

You will typically see two download options. We recommend choosing the LTS, or Long Term Support version, as it is the most stable. Click the button to download the installer for your operating system.

Once the download is complete, run the installer. The process is straightforward. You can generally accept the default settings and click through the prompts until it is finished.

Now, you need to verify that Node.js and npx are installed correctly. Open your terminal or PowerShell.

Type the following command to check the version:

node --version

If you see a version number printed out, you are good to go. This means you now have access to npx, which allows you to execute Node.js packages without installing them globally.

Now, you are ready to run the MCP Inspector. In your terminal, type the following command:

npx @modelcontextprotocol/inspector

When you run this for the first time, it may ask for permission to download the package. Type ‘y’ if prompted.

Once the command executes, you will see a message indicating that the MCP Inspector is running, usually on localhost:6277.

You can now open that link in your browser. You will see the Inspector interface, ready for you to connect your Python MCP servers. In the upcoming lessons, you will use this tool extensively to test the tools, resources, and prompts you build.