book-to-skill

book-to-skill is an open-source AI developer tool that converts a book or document into a structured, on-demand skill for coding agents, aimed at developers who want to give Claude Code, GitHub Copilot CLI, or Amp an author's expertise without bloating the context window. Rather than producing a summary, it distills named frameworks, mental models, decision rules, and anti-patterns — what the project calls the author's toolkit — into a compact SKILL.md file that the agent loads only when a topic is actually relevant. It accepts PDF, EPUB, DOCX, HTML, Markdown, RTF, and common e-reader formats, and extraction runs locally with standard-library fallbacks, so books aren't uploaded to a third party. Because it loads only the relevant chapters on demand, token cost scales with the scope of your query instead of the length of the document. The tool speaks the open Agent Skills standard, which is what lets the same generated skill work across multiple agents instead of locking you into one. It is MIT-licensed and dependency-light, installs via pip, and has drawn 6,000+ GitHub stars and a multilingual contributor community. For engineers who want their agents to reason with a specific book's methodology while keeping context lean and data private, book-to-skill is a focused local extraction tool rather than a hosted summarizer.

Turning books and documents into compact, on-demand skills for coding agents, for developers who want author-level methodology without context bloat.