MCP-FS (Model Context Protocol – File System)

A prototype toolhost that enables LLMs to perform basic file system operations through MCP over HTTP.

Features

• createFile – Create new files
• readFile – Read file content
• updateFile – Overwrite file content
• appendToFile – Append text to a file
• deleteFile – Delete files
• listFiles – List contents of a directory
• describeServer – Returns server tool list & usage guide

Architecture

• Server: Exposes tools via MCP HTTP transport using Express and @modelcontextprotocol/sdk.
• Client: Connects to MCP, fetches tool list, routes user input to an LLM (e.g., Ollama), and invokes tools based on LLM output.

Install & Run Ollama

MCP-FS uses Ollama to run local LLMs and respond to file system tool requests.
Ollama allows you to run models like mistral, llama2, codellama, and others locally via a simple API.
Official GitHub: ollama/ollama

Install Ollama

• macOS: Download

• Windows: Download

• Linux:

  curl -fsSL https://ollama.com/install.sh | sh

Start Ollama Server

ollama serve

By default, Ollama runs at http://localhost:11434.

Pull Required Model

ollama pull mistral-nemo

Check Installation

Should list available models (confirming Ollama is up and the model is installed):

curl http://localhost:11434/api/tags

Installation & Setup Project

Clone the repo

git clone https://github.com/your-username/mcp-fs.git
cd mcp-fs

Install dependencies

npm install

Start the MCP Server

npm run server

Start the MCP Client

npm run client

Resources

• Ollama Website: https://ollama.com
• Ollama GitHub: https://github.com/ollama/ollama
• Model Library: https://ollama.com/library