wtf is book-to-skill?

This project converts technical books and documentation into a format that AI coding assistants can reference while you work. Instead of searching through a long PDF every time you forget something, the tool turns the book into a structured skill file that your assistant loads on demand. You can use it with Claude Code, GitHub Copilot CLI, or Amp. The motivation is straightforward. You read a great technical book once, then three months later you cannot remember what was in chapter seven. Searching the PDF gives you page numbers, not answers. Asking an AI about the book produces guesses, since it does not actually have the content. This tool addresses that gap by extracting the book's structure into small files covering mental models, chapter summaries, a glossary, design patterns, and quick reference tables. To use it, you install the tool with pip, run a setup command, and then point it at a file or folder. It supports several formats including PDF, EPUB, DOCX, Markdown, HTML, RTF, and MOBI. For PDFs with lots of code or tables, an additional package called docling is recommended. Scanned PDFs require an OCR step first. The processing happens locally on your machine, so your files are never uploaded anywhere. The output is a collection of files. The main skill file contains core concepts and a chapter index, costing roughly 4,000 tokens. Individual chapter files sit alongside it and load only when you ask about a specific topic, which keeps the assistant's memory usage small. The project claims this approach uses 24 to 51 times fewer tokens than pasting the entire book into a conversation. Beyond books, you can feed it internal documentation, design system guidelines, research papers, or technical standards. The README notes the output is structured notes rather than a reproduction of the original text, and reminds users not to redistribute generated skills of copyrighted works. The full README is longer than what was shown.