TODO MCP CLI & Server

You run a minimal Model Context Protocol (MCP) server for a to-do list app. It exposes a server you can talk to via HTTP for tool discovery and JSON-RPC to manage tasks, plus a CLI client you can use to interact with the MCP server using natural language prompts. This setup lets you create, list, and complete tasks in a structured, extensible way while keeping logic on the server side.

How to use

Start the HTTP MCP server and connect a compatible MCP client to it. You can discover available tools through the server’s /tools endpoint and perform task operations by issuing JSON-RPC calls via the /rpc interface. Use the CLI client to input natural language requests like creating a new task, listing tasks, or marking tasks as done. The server handles task IDs and the underlying data changes, while the CLI translates your requests into the appropriate tool invocations.

How to install

# Prerequisites
Python 3.10+
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Server configuration
OPENAI_API_KEY=your_openai_api_key
MCP_SERVER=http://127.0.0.1:8000

# Run the server
python -m server.main

Additional notes

To run locally you will start a FastAPI server that serves the /tools and /rpc endpoints. Open a browser to http://127.0.0.1:8000/docs to inspect interactive API docs and test tool discovery and RPC calls. The CLI client prompts you in natural language and prints the tool invocations and LLM responses as it coordinates with the MCP server.

Available tools

add_task

Create a new task with a title, content, and optional due date. Returns the created task ID and metadata.

list_tasks

Retrieve all tasks with their status and details.

complete_task

Mark a specific task as completed by its ID.