GitHub radar
Book to Skill: Load Any Book into Claude Code (link)
Create a Claude Code skill that lets your agent answer any question based on information found in a book, PDF, or directory of documents. At 24–51x less token cost than reading the entire book into context, this is far more efficient!
Book to Skill creates a Claude Code skill, GitHub Copilot CLI, Amp, or Codex skill out of any technical book, directory of documents, or collection of files. Rather than reading an entire book into context at once, it identifies relevant frameworks, heuristics, anti-patterns, and chapter files which are then read into context on demand (when asked about that specific subject). Authors: 24–51x fewer tokens required to achieve same results as blasting the entire book into context. Input formats: PDF, EPUB, DOCX, MD, HTML, RTF, MOBI. Contains: SKILL.md (~4,000 tokens) (an index), chapter files (~1,000 tokens/file), glossary, patterns reference, cheatsheet. Can be used for non-books too: architecture docs, style guides, research paper sets, RFCs, compliance standards, etc. Anything you can use as a reference that you’d like to memorize.
Why a vibe-coder should care
Never get incorrect answers about what the book says again. Ask /book-slug topic and the agent will read the appropriate chapter for you. Install once per book, use in all Claude Code sessions.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install the book-to-skill agent skill from https://github.com/virgiliojr94/book-to-skill into Claude Code — clone it into ~/.claude/skills/book-to-skill per the README, then run /book-to-skill on a PDF file I'll point you to, and show me the structure of the generated skill.
Runs on a regular laptop — Python (for book conversion) and Claude Code required.
Open on GitHub▌ More finds