Quick Start Guide

This guide will walk you through launching MCPJam Inspector and connecting to a sample MCP server in just a few minutes.

Step 1: Launch the Inspector

Open your terminal and run the following command to start the Inspector using npx:

npx @mcpjam/inspector@latest

This will download the latest version and start the web server. Your default browser should open to http://localhost:3001. If it doesn't, you can navigate there manually.

Step 2: Explore the Interface

You are now on the MCP Servers tab. The interface is organized as follows:

  • Sidebar (Left): Navigate between different sections like the Playground, Tools, Resources, and Settings.
  • Main Content Area: This is where you'll interact with servers, view tools, and chat with models.

Step 3: Add Your First MCP Server

We'll connect to a sample MCP server that exposes a variety of example tools. This server runs locally on your machine using STDIO.

  1. Click the + Add Server button.

  2. A modal will appear. Fill in the following details:

    • Server Name: everything-server
    • Connection Type: Select STDIO.
    • Command: Enter npx -y @modelcontextprotocol/server-everything

  3. Click Add Server.

The Inspector will now start the server process and establish a connection. You should see a new card for everything-server with a green "Connected" status.

Step 4: Inspect the Server's Tools

Now that you're connected, let's see what the server can do.

  1. In the left sidebar, click on Tools.
  2. The everything-server should be automatically selected in the top bar.
  3. The sidebar on the Tools screen now shows a list of all tools exposed by the server, such as get_weather, get_current_time, and file_read.
  4. Click on any tool, like get_weather, to see its description and input parameters.

Step 5: Test the Tools in the Playground

Finally, let's interact with the server using natural language in the Playground.

  1. In the left sidebar, click on Playground.
  2. If you have configured an API key for a provider like OpenAI or Anthropic in the Settings tab, you can start chatting. If not, the input will be disabled.

  3. Type a message that should invoke one of the server's tools, such as:

    What is the weather in San Francisco?
  4. Press Enter. The LLM will process your request, identify the need to use the get_weather tool, call the MCP server, and return the result.

What's Next?

You've successfully launched the Inspector, connected to an MCP server, and tested its tools! From here, you can:

  • Explore the Usage Guide to learn about all the features in detail.
  • Learn how to configure servers using a mcp.json file.
  • Try one of our Tutorials to build your own MCP server.