Show HN: A Claude Code skill to analyze your chess games
The project is a set of Claude Code skills that turn one of your chess games into a post-mortem you can actually read: plain-language explanations of your mistakes, checked against Stockfish, and a narrated video of the whole game. The post hit the front page of Hacker News.
The demo is a real game of the author's — a 15+10 rapid on lichess (ID 5KmlrdyT), playing White in an Open Sicilian against a Najdorf at around 1700. During the game he recorded himself thinking aloud in French with a plain voice recorder. Afterwards he gave Claude two things: the lichess link and the mp3. Claude transcribed the audio locally with whisper.cpp, then used the clock times in the PGN to match each sentence to the move he was deciding at that moment — so when the video says "at move eight you asked yourself whether the bishop belongs on c4 or e2," that is his own question answered with the engine. Everything the video shows (the transcript, the Stockfish sweep, the annotated PGN, the storyboard) is in examples/game-010/. The video runs 6:49 with English narration and subtitles and is a 10 MB mp4; the embedded player requires a GitHub login.
The motivation: a Stockfish analysis gives you numbers and lines, which are not fun to read and often do not tell you why — say -1.8 after your move, plus a 12-move variation you would never find at the board. Claude can drive Stockfish itself, asking the engine the questions a human would ask ("why not the move I played?", "what if Black simply takes?"), keep asking until the answer makes sense, and write it up in human terms. Because it knows what the player was thinking during the game, it can refute the actual reasoning instead of a guessed one, and afterwards you can keep asking your own questions about any position, with the engine checking them before it answers.
On fit: it is probably for you if you play slow or rapid games and want to learn from them rather than just see where the eval bar dropped, if you are fine running Claude Code on your machine, and if you are willing to take notes or talk during the game (optional, but that is what makes the result interesting). It is probably not for you if you want a quick check of a blitz game, which lichess or chess.com does in seconds. A full run (sweep, investigations, annotations, video) takes around an hour of processing, so it should be used on games where you took time to think, not on every game you play.
The post strongly recommends passing written notes or an audio recording covering the whole game: without them you get a good engine explanation, but with them you get a review of your reasoning — which fears were justified, which plans were sound, and which questions you asked yourself and what the answer was. The author also flags hallucinations as a remaining risk, noting that every claim is checked against Stockfish (the excerpt truncates here).