Mcpfilesystem Client

🚀 Usage Guide for LLM Application

This guide provides a step - by - step process for setting up and running an application that interacts with a large language model (LLM) through a local server.

🚀 Quick Start

Step 1: Launch the Local LLM Server

Start a large language model (LLM) compatible with the OpenAI SDK on your local server. You can use Jan to achieve this.

Step 2: Modify the .env File

Edit the .env file and add the following information:

• The name of the model you are using.
• The prompt text.
• The base URL of the server (which is set by default).

Step 3: Run the Program

Execute the command node app.js to start the application.

⚠️ Important Notes

Access Permission Limitation

The program can only access the specified directory, which is the files directory by default. If you need to change it, adjust the server MCP section in app.js:

// EJ ln.20
const transport = new StdioClientTransport({
    command: "npx",
    args: [
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "ruta",
        "otra ruta",
        ...
    ],
});

Default Model Issue

By default, the model specified in the .env file may not work properly because it does not support the tool - calling function. Make sure to select a model that supports tool calls.