Running the MCP Server with Inspector

Starting the MCP Inspector

Below inspector can be used to monitor and debug your MCP server. To start the MCP Inspector, run the following command in your terminal:

npm install -g @modelcontextprotocol/inspector
mcp-inspector

This will launch the MCP Inspector in your default web browser.

Related MCP server: Code Sandbox MCP

Running the MCP Server

In a separate terminal, navigate to the directory where your MCP server code is located and start the server:

source venv/bin/activate  # Activate your virtual environment if you have one
python3 main.py

Connecting to the MCP Server

Once the MCP Inspector is running, you can connect to your MCP server as follows:

STDIO Transport

1. In the MCP Inspector, select Transport Type (by default STDIO)

2. In the Command field, enter the command to start your MCP server (e.g., python3).

3. In the Arguments field, enter the path to your MCP server script (e.g., main.py).

4. Click the "Connect" button to establish a connection between the MCP Inspector and your MCP server.

Streamable HTTP Transport

1. In the MCP Inspector, select Transport Type (Streamable HTTP).

2. In the URL field, enter the URL where your MCP server is running (e.g., http://localhost:8005/mcp).

3. Click the "Connect" button to establish a connection between the MCP Inspector and your MCP server.

You should now see the MCP server's logs and interactions in the MCP Inspector interface, allowing you to monitor and debug your MCP server effectively.

List Tools in the tools tab and Test. You can see the available tools and their descriptions in the MCP Inspector's Tools tab. You can also test the tools by sending requests to your MCP server and observing the responses in the MCP Inspector.