📌Overview

Purpose: This framework aims to facilitate the integration of AI models with various tools through a unified interface using the Model Context Protocol (MCP) and Ollama.

Overview: The TypeScript MCP Agent with Ollama integration provides a robust platform for AI models to perform file system operations and web research. It allows easy configuration and testing through a standalone demo mode, ensuring fluid interaction with the tools via a chat interface.

Key Features:

Multi-Server Support: Supports multiple MCP servers (tested with uvx and npx), enabling flexibility in deployment.
Tool Operations: Built-in functionality for file system actions and web research, enhancing the capabilities of AI models.
Simple Configuration: Contains an easy setup process utilizing mcp-config.json for tool and model configurations, making it user-friendly.
Interactive Chat Interface: Facilitates seamless interaction with Ollama, supporting any compatible tools for a more engaging user experience.
Standalone Demo Mode: Allows testing of web and filesystem tools independently of a language model, providing a straightforward evaluation environment.

TypeScript MCP Agent with Ollama Integration

This project demonstrates integration between Model Context Protocol (MCP) servers and Ollama, allowing AI models to interact with various tools through a unified interface.

✨ Features

Supports multiple MCP servers (both uvx and npx tested)
Built-in support for file system operations and web research
Easy configuration through mcp-config.json similar to claude_desktop_config.json
Interactive chat interface with Ollama integration that supports any tools
Standalone demo mode for testing web and filesystem tools without an LLM

🚀 Getting Started

Prerequisites

Node.js (version 18 or higher)
Ollama installed and running
Install the MCP tools globally that you want to use:

# For filesystem operations
npm install -g @modelcontextprotocol/server-filesystem

# For web research
npm install -g @mzxrai/mcp-webresearch

Clone and install

git clone https://github.com/ausboss/mcp-ollama-agent.git
cd mcp-ollama-agent
npm install

Configure tools and Ollama model in `mcp-config.json`

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["@modelcontextprotocol/server-filesystem", "./"]
    },
    "webresearch": {
      "command": "npx",
      "args": ["-y", "@mzxrai/mcp-webresearch"]
    }
  },
  "ollama": {
    "host": "http://localhost:11434",
    "model": "qwen2.5:latest"
  }
}

Run the demo (without an LLM)

npx tsx ./src/demo.ts

Start the chat interface with Ollama

npm start

⚙️ Configuration

Add any MCP-compatible server to the mcpServers section
Configure host and model for Ollama (model must support function calling)
Supports both Python (uvx) and Node.js (npx) MCP servers

💡 Example Usage

This example uses the model qwen2.5:latest

Chat started. Type "exit" to end the conversation.
You: can you use your list directory tool to see whats in test-directory then use your read file tool to read it to me?
Model is using tools to help answer...
Using tool: list_directory
With arguments: { path: 'test-directory' }
Tool result: [ { type: 'text', text: '[FILE] test.txt' } ]
Assistant:
Model is using tools to help answer...
Using tool: read_file
With arguments: { path: 'test-directory/test.txt' }
Tool result: [ { type: 'text', text: 'rosebud' } ]
Assistant: The content of the file `test.txt` in the `test-directory` is:
rosebud
You: thanks
Assistant: You're welcome! If you have any other requests or need further assistance, feel free to ask.

System Prompts

Customize the system prompt in ChatManager.ts if local models need help with tool selection.

🤝 Contributing

Contributions are welcome! Feel free to submit issues or pull requests.

mcp-ollama-agent