Chesster is a conversational AI Chess teacher. Given a PGN dump of a game or sequence of moves, Chesster will walk you through interesting moments, highlight successes and mistakes, and simulate alternate scenarios.
Chesster will also attend as you play a game against the computer (currently the Stockfish engine), answering questions and identifying opportunities to learn.
demo.mp4
Currently, the only mode of interaction with Chesster is via natural language conversation. Moves are made by describing them or specifying them with algebraic or UCI notation. Chesster will modify the board and take other actions based on your queries.
Chesster includes an application server and a separate Langserve server for LLM orchestration. They can be built and launched with
docker-compose build base
docker-compose build langserver app
OPENAI_API_KEY=... docker-compose up
Alternatively, you can run locally with
OPENAI_API_KEY=... make start
make unit_tests