Claude Code skill analyzes chess games with plain-language explanations and video

AI-Powered Chess Analysis

A new Claude Code skill has been developed to analyze chess games, offering a post-mortem that includes plain-language explanations of player mistakes. The analysis is checked against the Stockfish chess engine to ensure accuracy. The tool also produces a narrated video of the entire game, detailing key moments and decisions.

Integrating Human Thought and Engine Power

The skill can integrate a player's in-game thoughts, recorded via audio or notes, to provide more personalized feedback. Claude transcribes audio using whisper.cpp and matches sentences to specific moves based on PGN clock times. This allows the system to refute a player's actual reasoning with engine-backed explanations, rather than generic analysis.

Beyond Standard Engine Analysis

Unlike typical Stockfish analyses that provide numerical evaluations and complex variations, this Claude skill aims to answer human-like questions about moves and positions. It iteratively queries the engine until explanations are clear and understandable. This approach makes engine analysis more accessible and actionable for players seeking to learn from their games.

Use Cases and Limitations

This tool is suitable for analyzing slow or rapid games where players want in-depth learning beyond simple evaluation bar drops. It requires running Claude Code locally and benefits significantly from player input like recorded thoughts. A full analysis, including sweep, investigations, annotations, and video generation, takes approximately one hour, making it less ideal for quick checks of blitz games.