mcp server
duck-goose
Experiment with model context protocol (MCP)
Set up
Runs on your computer. Pick your agent and operating system.
This project does not publish an install method we can turn into a config. Follow the README instead.
README
synced Oct 23, 2026duck_goose
An experiment with model context protocol (MCP).
Requirements
- Install
uv(environment management for Python) - Install
ollama(e.g.brew install ollama) (model management and server for llm model) - See below for instructions.
Demonstrations goal
The goal of all demonstrations here involves asking an LLM to create a database and single table with records "duck", "duck", "goose". We ask the LLM to perform this work using MCP or local Tools. This allows the LLM to agentically achieve this work without specifying explicit programmatic instructions to accomplish this work.
Running the demonstrations
We use poethepoet to create declarative tasks which run the demonstrations found here.
You may list these by using uv run poe.
Run these demo by using uv run poe <task name>.
The following tasks exist:
uv run poe local_tool_demo: Run a local tool-based LLM example.uv run poe mcp_demo: Run an mcp-based tool LLM example.