Chess LLM Tutor

An interactive chess tutor that uses LLMs to provide real-time coaching. Built as my university thesis.

How it works

• Play chess against a Maia engine (human-like neural network chess engine)
• Chat naturally — ask about positions, strategy, mistakes, or next moves
• Intent classification routes your message (analysis request, move command, general question)
• LLM generates responses grounded in the current board state via structured prompts

Stack

• Backend: Django + python-chess + Maia engine
• Frontend: Chessboard.js + vanilla JS
• LLM: Claude via structured prompt pipelines

Architecture

views.py          → API endpoints (chat, moves, board state)
ChessLogic.py     → Board management, move validation, Maia integration
PromptMaker.py    → Context-aware prompt construction from board state
intent.py         → Classifies user messages into action types
maia_engine.py    → Human-like chess engine for adaptive difficulty