Show HN: A Claude Code skill to analyze your chess games
An open-source Claude Code skill uses Stockfish and local transcription to turn chess games and think-aloud audio into plain-language post-mortem videos and annotated PGNs.
summary is enough While a heavy and costly workflow, it offers a neat architectural pattern for combining Claude Code subagents with local CLI tools like Stockfish and Whisper.
summary
The project provides a set of Claude Code skills that automate chess game post-mortems by pairing Stockfish engine evaluations with local audio transcriptions of a player's thought process. Subagents iteratively interrogate Stockfish to explain mistakes in plain language rather than raw evaluation numbers. The final output includes an annotated PGN, an interactive HTML board, and a narrated video complete with text-to-speech and subtitles. Though resource-intensive and slow, it attempts to bridge the gap between raw engine output and human comprehension.
key points
- Subagents iteratively query Stockfish with human-like questions to ground LLM commentary in verified engine lines.
- The tool integrates local audio transcription via whisper.cpp matched to PGN clock timestamps to review a player's actual in-game reasoning.
- It compiles multi-modal deliverables including layered annotated PGNs, standalone HTML review boards, and fully narrated video files.
who should read it
Software developers and chess enthusiasts interested in building custom AI workflows or automating game post-mortems.
discussion mixed
consensus
Commenters find the integration of Stockfish with LLMs compelling for natural-language commentary, though many debate whether LLMs add genuine value over standard engines given high token costs and hallucination risks.
pushback
Skeptics argue that relying on LLMs for chess analysis is costly, lazy, and fundamentally prone to reasoning errors compared to traditional tools.
You only need to spend a couple of minutes talking to an LLM about chess to realize it has no understanding of the game. It will produce reasonable sounding explanations for things that are completely wrong. It will make illegal moves. It will hallucinate pieces on squares even when the entire PGN is in the context. There are no prompts you can give it to fix this. I don't really understand how you could create this skill without noticing this.
It highlights the core pitfall of LLMs in chess, noting that models frequently struggle with basic game rules and hallucinate positions despite having full context.
Generated by an LLM from the article and thread. Check the sources before relying on it.