Basketball RL MCP Chatbot

An AI chatbot powered by an MCP (Model Context Protocol) server that answers questions about my Reinforcement Learning Graduate Project. Unlike a RAG system that searches documents, this uses structured tool functions that Claude dynamically selects and calls to retrieve precise project data.

Live Demo

https://basketballmcp-znyihwsjejzsuudrrhyzkl.streamlit.app/

How it works

• server.py defines MCP tools exposing basketball RL training data
• data.py contains structured project data including training rounds, hyperparameters, and infrastructure details
• utils.py handles tool calling and Claude API communication
• Claude figures out which tool to call based on the user's question and returns accurate data

Tech Stack

• Anthropic API (Claude)
• MCP / FastMCP
• Streamlit
• Python

Related Project

This chatbot is built on top of my Multi-Agent RL Basketball Simulation: https://github.com/xb2101/basketball-defender-rl

How to run locally

1. Clone the repo
2. Install dependencies: pip install -r requirements.txt
3. Set your Anthropic API key: export ANTHROPIC_API_KEY=your-key-here
4. Run the Streamlit app: streamlit run streamlit_app.py
5. Or run the terminal version: python client.py