35.2K ·GitHub
What is this?
book-to-skill turns a technical book PDF (or a folder/glob of documents) into a structured 'skill' that coding agents like Claude Code, GitHub Copilot CLI, Amp, or Hermes Agent can load on demand to answer questions from the real content instead of hallucinating.
book-to-skill turns a technical book PDF (or a folder/glob of documents) into a structured 'skill' that coding agents like Claude Code, GitHub Copilot CLI, Amp, or Hermes Agent can load on demand to answer questions from the real content instead of hallucinating.Is this a converter tool or a finished app?It is a two-part tool: a deterministic Python extractor that pulls clean text out of your source files, plus a spec-driven generator your agent runs to turn that text into a SKILL.md-based skill — there's no separate GUI or server.
It is a two-part tool: a deterministic Python extractor that pulls clean text out of your source files, plus a spec-driven generator your agent runs to turn that text into a SKILL.md-based skill — there's no separate GUI or server.How do I turn a book into a skill?Point the /book-to-skill slash command at a file, folder, or glob inside a supported agent, e.g. /book-to-skill ./my-book.pdf; the output (SKILL.md plus per-chapter files) lands in that agent's skills directory and you then query it with /your-book-slug.
Point the /book-to-skill slash command at a file, folder, or glob inside a supported agent, e.g. /book-to-skill ./my-book.pdf; the output (SKILL.md plus per-chapter files) lands in that agent's skills directory and you then query it with /your-book-slug.What do I need installed to extract PDFs?Python 3 is required; for PDFs it uses pdftotext (poppler), pypdf, pdfminer.six or docling depending on whether the book is text-heavy or technical, plus similar per-format tools for EPUB, DOCX, HTML and RTF. Running python3 scripts/extract.py --check shows what's missing and the exact install command.
Python 3 is required; for PDFs it uses pdftotext (poppler), pypdf, pdfminer.six or docling depending on whether the book is text-heavy or technical, plus similar per-format tools for EPUB, DOCX, HTML and RTF. Running python3 scripts/extract.py --check shows what's missing and the exact install command.Does it work on scanned PDFs?No — a scanned PDF that's just page images with no text layer has nothing to extract, so the tool stops immediately with an explanation instead of generating an empty skill; you need to OCR it yourself first.
No — a scanned PDF that's just page images with no text layer has nothing to extract, so the tool stops immediately with an explanation instead of generating an empty skill; you need to OCR it yourself first.Does my book content get uploaded anywhere?No, extraction and analysis run locally on your machine and the tool never uploads your files; if your agent's underlying model is cloud-based, only the text you feed it follows that provider's normal terms, same as any prompt.
No, extraction and analysis run locally on your machine and the tool never uploads your files; if your agent's underlying model is cloud-based, only the text you feed it follows that provider's normal terms, same as any prompt.Can I share a generated skill of a book I bought?The MIT license covers only the converter code, not any book you process; redistributing a generated skill of a copyrighted third-party book can infringe the rights holder, so keep those private.
The MIT license covers only the converter code, not any book you process; redistributing a generated skill of a copyrighted third-party book can infringe the rights holder, so keep those private.What other inputs besides books can it turn into a skill?Besides books, it works on internal docs like runbooks and ADRs, brand/design guideline PDFs, clusters of research papers, and specs or RFCs — basically anything you re-read often enough to wish you'd memorized it.
Besides books, it works on internal docs like runbooks and ADRs, brand/design guideline PDFs, clusters of research papers, and specs or RFCs — basically anything you re-read often enough to wish you'd memorized it.How much does the skill save on tokens versus dumping the PDF into context?Measured on real books, querying the generated skill uses 24x-51x fewer tokens than feeding the whole book into context for one question, since chapter files only load on demand.
Measured on real books, querying the generated skill uses 24x-51x fewer tokens than feeding the whole book into context for one question, since chapter files only load on demand.Can I update a skill after adding new chapters or sources?Yes, it has update and fold-in modes that merge new material into an existing skill instead of regenerating it from scratch.
Yes, it has update and fold-in modes that merge new material into an existing skill instead of regenerating it from scratch.Which agents can actually run the generated skill?Any host implementing the open Agent Skills standard — currently GitHub Copilot CLI, Amp, Claude Code, and Hermes Agent — since they all read the same SKILL.md format.
Any host implementing the open Agent Skills standard — currently GitHub Copilot CLI, Amp, Claude Code, and Hermes Agent — since they all read the same SKILL.md format.What are similar projects?docling, marker, qiaomu-anything-to-notebooklm, ai-aug-edu, OpenKB, patent-disclosure-skill, chubbyskills, SenseNova-Skills
docling, marker, qiaomu-anything-to-notebooklm, ai-aug-edu, OpenKB, patent-disclosure-skill, chubbyskills, SenseNova-SkillsWhich topics is it in?docx, epub, knowledge-management, context-engineering, document-processing, knowledge-base, pdf, rag, html, markdown, agent-skills, claude-code
docx, epub, knowledge-management, context-engineering, document-processing, knowledge-base, pdf, rag, html, markdown, agent-skills, claude-code