Ollama MCP Example

Based on projects:

• https://ollama.com/
• https://github.com/open-webui/open-webui
• https://github.com/flexnst/ollama-mcp-bridge
• https://github.com/sparfenyuk/mcp-proxy
• https://github.com/modelcontextprotocol/servers/blob/main/src/fetch/README.md

Setup

1. Clone and start docker compose environment:

git clone https://github.com/flexnst/ollama-mcp-example.git
cd ollama-mcp-example
make init

2. Wait for docker images pulled and built
3. Go to Open-WebUI interface: http://localhost:3000
4. Select and download your LLM models (see on https://ollama.com/search?c=tools)

make ollama-cli
ollama pull qwen2.5:7b # for example

5. Good luck with your AI experiments!

Available commands:

• make init - Init new docker compose environment
• make up - Up environment
• make down - Stop enviroment
• make ollama-cli - Docker ollama enviroment

Tools available out of the box:

fetch

This tool enables LLMs to retrieve and process content from web pages, converting HTML to markdown for easier consumption.

How to add my tools?

compose.yml

  mcp-proxy:
    container_name: mcp-proxy
    image: mcp-proxy:latest
    build:
      context: mcp-proxy
      dockerfile: Dockerfile
    command: |
      --pass-environment --host=0.0.0.0 --port=8096 --allow-origin=*
      --transport=sse
      -- uvx mcp-server-fetch --ignore-robots-txt --user-agent='${FETCH_USER_AGENT}'

You can add your tools using uvx, for additional info see: https://github.com/sparfenyuk/mcp-proxy#22-example-usage

Where can I see the available tools (mcp servers)?

https://github.com/modelcontextprotocol/servers

License

MIT.