ollama-mcp-example

Overview

This repository provides a straightforward example of implementing MCP (Model Context Protocol) with Ollama. It demonstrates how to create an MCP server that exposes tools over a network and an MCP client that connects to these servers to make them available to an LLM. The example includes clear implementation of both server and client sides, showing how tool calls are routed through MCP instead of executing locally.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• get_weatherget_weather(city: str) -> str

  Gets weather information for a specified city

  Try:What's the weather like in New York?Try:Is it sunny in London today?

Comparable tools

Installation

git clone https://github.com/kirillsaidov/ollama-mcp-example.git
cd ollama-mcp-example

python3 -m venv venv
./venv/bin/pip install -r requirements.txt

# Start MCP server
./venv/bin/python mcp_server.py

# Run MCP client
./venv/bin/python mcp_client.py

{
  "mcpServers": {
    "ollama-mcp": {
      "command": "python",
      "args": ["mcp_server.py"],
      "env": {
        "PYTHONPATH": "."
      }
    }
  }
}

FAQ

What is the difference between MCP and regular function calling?

With MCP, you don't need to implement, define, or execute the tools yourself. MCP servers handle that. Most importantly, they are reusable and model-agnostic.

How can I add my own functions to this example?

Simply add your own decorated function to mcp_server.py using the @mcp.tool() decorator.

Compare ollama-mcp-example with