book-to-skill for Claude Code: What It Does and How to Install It

Makes your coding agent turn a book or document folder into a new skill of frameworks, principles and techniques.

Installs: 8.9K; on skills.sh
Stars: 36.7K; on GitHub
Updated: Oct 2026; last commit

You run /book-to-skill with a file, folder or glob. A bundled Python script extracts the text, the agent asks whether the content is technical or text-heavy, shows a token estimate and waits for your go-ahead. It then writes a skill with SKILL.md, chapter summaries, a glossary, patterns and a cheatsheet. It also has analyze-only, generate-from-analysis and update modes.

It needs Python and a shell. It handles PDF, EPUB, DOCX, HTML, Markdown, text and RTF, and MOBI or AZW with Calibre. Optional extractors such as docling are offered when a better one is missing. The README says to keep skills made from copyrighted books private.

Good for

People who want to study a book through their coding agent
Readers who want an author's frameworks applied while they work
Teams that want a reusable knowledge base from a folder of documents

Install book-to-skill

Claude Code

Copy this prompt.

Install the book-to-skill skill in Claude Code. Run this command:

npx skills add virgiliojr94/book-to-skill

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

The skills CLI route is the README one-liner. The git clone routes are the manual installs the README and docs/install.md give. The standalone pip CLI in docs/install.md installs only the text extractor and does not register the skill. The skill runs Python and writes files, so there is no Claude app upload route. Install it only from virgiliojr94/book-to-skill; see the security notice in the permissions section.

From the project README. Source.

What's in the skill

When it loads

Converts books and documents (PDF, EPUB, DOCX, HTML, Markdown, plain text, RTF, MOBI/AZW with Calibre) into structured agent skills, extracting frameworks, mental models, principles, techniques, and anti-patterns. Use when the user wants to study a document through GitHub Copilot CLI, Amp, Claude Code, Hermes Agent, OpenCode, or OpenClaw, apply an author's frameworks while working, or build a reusable knowledge base from a file.

Sections in SKILL.md

Philosophy
Modes of Operation
1. Full Conversion (Default)
2. Analyze Only
3. Generate from Prior Analysis
4. Update / Fold-in (Existing Skill)
Skill Locations
Step 0 — Out-of-scope check
Step 1 — Validate input
Step 1.5 — Identify content type
Step 2 — Extract text from the source documents
Step 2.5 — Pre-flight cost estimate
Step 2.6 — REPL-style access for large books (> 50k tokens)
Step 3 — Analyze book structure
Extraction Report — <Title>
Author's Core Frameworks

Key facts

Author: virgiliojr94 (LICENSE.md copyright holder)
License: MIT (LICENSE.md and README badge)
Installs on skills.sh: 8,907 (Nov 2, 2026)
GitHub stars: 36,701 for the repo (Nov 2, 2026)
Last repo commit: Oct 28, 2026
Audits on skills.sh: Gen Agent Trust Hub: pass, Socket: pass, Snyk: pass
What it includes: The repo has 132 files, including 60 tests and 18 docs; the skill uses SKILL.md, scripts/extract.py and tools/
Needs: python3; optional extractors; Calibre for MOBI/AZW
Works in: GitHub Copilot CLI, Amp, Claude Code, Codex, Hermes Agent, OpenCode, OpenClaw

What it includes

Runs Steps 0 to 9 and writes the whole skill
Output: SKILL.md, chapters/, glossary.md, patterns.md and cheatsheet.md.
Try: “/book-to-skill ~/books/thinking-fast-and-slow.pdf”
Runs code · Full Conversion (default)
Runs Steps 0 to 3, then stops with an extraction report
Say "analyze" or "just extract" to use it.
Runs code · Analyze Only
Skips Steps 0 to 3 and builds the skill from your analysis notes
Write · Generate from Prior Analysis
Merges new sources into an existing skill's chapters, indexes and glossary
Runs code · Update / Fold-in
Turns the files into full_text.txt and metadata.json in a work folder
Run with --mode and --install-missing ask; --check prints which extractors are installed.
Runs code · Extractor: scripts/extract.py
Step 2.5 shows tokens and time and waits for your confirmation
Read · Pre-flight cost estimate
Step 9.5 runs tools/scan_generated_skill.py and stops on findings
Runs code · Scan the generated skill
Step 10 symlinks into ~/.claude/skills and removes this run's work folder
Delete · Link and clean up
Step 11 can run git init and gh repo create, private by default
It asks for your go-ahead first, and asks private or public as a separate question.
Runs code · Publish to GitHub (optional)

From the official docs

Permissions

Medium risk

The skill runs Python, writes and links files in your skills folders, and can offer to publish the result to GitHub. Install it only from the official repo and keep skills made from copyrighted books private.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

A malicious re-upload exists: Leutenegger/book-to-skill
SECURITY-NOTICE.md (dated 2026-08-17) says the only official repo is virgiliojr94/book-to-skill and that Leutenegger/book-to-skill is not affiliated with it. The notice says the maintainer reviewed the re-upload's code and found behavior absent from the official project: disabling TLS certificate verification, sending host, system and repository metadata to an external Cloudflare Worker, enumerating local browser-extension storage for several cryptocurrency wallets and Ledger application data, archiving and uploading collected data to an external endpoint on macOS, and shipping a Windows ZIP/EXE payload the modified CLI can extract and launch. It tells anyone who ran it on a machine with wallet software to treat that wallet data as potentially compromised. Check the owner before you install. github.com
It can install Python packages
The extractor runs with --install-missing ask and prompts before installing a better extractor. In a non-interactive session it uses the fallback unless install mode is explicitly yes. github.com
It writes outside your project and deletes a folder
Step 10 runs ln -sfn to link the new skill from ~/.agents/skills into ~/.claude/skills. It then runs rm -rf "$WORKDIR", where SKILL.md says WORKDIR is the work directory this run reported (by default <tempdir>/book_skill_work-<pid>/, or the path in BOOK_SKILL_WORKDIR) and tells the agent never to delete a directory it did not create. Step 5 can also offer to migrate or overwrite an existing skill. github.com
The last step can publish to GitHub
Step 11 runs git init and gh repo create after you accept. Private is the default, and public needs the bare word "public". Keep skills from copyrighted books private. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only switch exists.
Talks to	GitHub (only if you accept the optional publish step)
Sign-in and keys	The skill names no key. Step 11 uses your signed-in gh CLI if you accept.
Scopes	Reads the documents you point it at Runs scripts/extract.py and tools/scan_generated_skill.py with Python Writes a skill folder under ~/.agents/skills and links it into ~/.claude/skills Removes the work folder it created Optional: creates a GitHub repo and pushes the generated skill

Questions

What does book-to-skill do?

It makes your coding agent convert a book or a set of documents into a skill. The skill holds the author's named frameworks, principles, techniques and anti-patterns, plus chapter summaries, so the agent can use them later.

How do I install it?

Use the install box at the top of this page. The README gives a skills CLI one-liner and a git clone into your skills folder. You need Python for the extractor.

Which files can it read?

PDF, EPUB, DOCX, text, Markdown, reStructuredText, AsciiDoc, HTML, RTF, and MOBI, AZW or AZW3 with Calibre. Plain text, Markdown, reStructuredText and AsciiDoc need no extra packages.

Does it install Python packages?

It can offer to. SKILL.md runs the extractor with --install-missing ask, which prompts you when a better extractor is missing, and non-interactive sessions fall back unless install mode is yes.

Is it safe?

It runs a Python script, writes files under your skills folder, creates a symlink and runs rm -rf on the work folder it reported for the run. Its optional last step can create a GitHub repo and push the result. Install it only from virgiliojr94/book-to-skill: the project's SECURITY-NOTICE.md (2026-08-17) says a re-upload at Leutenegger/book-to-skill disabled TLS verification and collected and uploaded system and crypto-wallet data. Read the full notice in the permissions section.

Can I publish a skill made from a book?

The README says skills from third-party copyrighted books should stay private. Public is offered only for your own writing, openly licensed text or material you may redistribute.

Does it work in the Claude app?

No upload route is listed, because the skill runs a Python script and writes files. Use it in a coding agent such as Claude Code.

Information

updated Nov 2, 2026

Repository: virgiliojr94/book-to-skill

virgiliojr94/book-to-skill skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 4

book-to-skill README · github.com
book-to-skill docs/install.md · github.com
book-to-skill SKILL.md · github.com
book-to-skill SECURITY-NOTICE.md · github.com