Ollama MCP Client

A Model Context Protocol (MCP) client for language models served through Ollama.

Features

• Multiple Model Support: Works with Qwen, Llama 3, Mistral, and Gemini models served via Ollama
• Streaming Responses: Real-time streaming of LLM responses
• Database Operations: Seamless integration with the Database MCP Server
• Natural Language Interface: Interact with databases using natural language

Installation

# Clone the repository
git clone <repository-url>

# Install the package
pip install -e .

Quick Start

# Basic usage
mcp-ollama --model llama3 --server http://localhost:8000 "List all tables in my database"

# Specify a different Ollama server
mcp-ollama --model qwen --ollama-api http://localhost:11434 --server http://localhost:8000 "Show me the schema for the users table"

Configuration

The client can be configured using command line arguments or a configuration file:

Command Line Arguments

• --model: The Ollama model to use (e.g., llama3, qwen, mistral, gemini)
• --server: URL of the Database MCP server
• --ollama-api: URL of the Ollama API server (default: http://localhost:11434)
• --stream: Enable/disable streaming responses (default: true)
• --config: Path to configuration file

Configuration File

Create a config.json file in your home directory:

{
  "model": "llama3",
  "server": "http://localhost:8000",
  "ollama_api": "http://localhost:11434",
  "stream": true
}

Example Usage

Natural Language Database Queries

# List tables
mcp-ollama "What tables do I have in my database?"

# Query data
mcp-ollama "Show me all users who registered in the last month"

# Create a table
mcp-ollama "Create a new table called products with columns for id, name, price, and description"

License

MIT