Book to Skill

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites — Mapped perimeter first: this listing installs exactly 19 files — the root SKILL.md, the whole book_to_skill/ Python package (17 files) and scripts/extract.py — under their original names; the rest of the repository (docs/, tests/, evals/, tools/, mkdocs/pyproject site and packaging tooling, READMEs) is repository infrastructure and is NOT installed. No account, key or payment is needed to install, and no script is run during installation. At use time the skill needs Python and runs its extraction script on documents you give it, on your machine: optional extractors for some formats are installed only after it explicitly asks you (--install-missing ask), MOBI/AZW3 go through Calibre if present, and before any generation the skill shows you a pre-flight token-cost estimate (Step 2.5) so you can stop there. The books you convert keep their own rights; this listing grants none. Install "Book to Skill" for me. It converts books and documents into structured agent skills: instead of a summary it extracts structure — named frameworks, actionable principles, step-by-step techniques, anti-patterns and the author's voice — and generates a new skill folder with a master SKILL.md, per-chapter files, a glossary, patterns and a cheatsheet. Its workflow asks what kind of content the sources hold (technical or text-heavy, which selects the extractor), runs the bundled Python extractor (scripts/extract.py, backed by the book_to_skill package in this same install) over the files you pass — PDF, EPUB, DOCX, HTML, Markdown, plain text, RTF, and MOBI/AZW/AZW3 through Calibre — presents the pre-flight token-cost estimate from the extraction metadata before generating anything, then analyses, asks the purpose, names and generates the skill, scans what it generated, and cleans up. It also offers analyze-only, generate-from-analysis and update/fold-in modes for existing skills, and an optional Step 11 that publishes the generated skill to GitHub only when you ask for it. This listing's install perimeter is mapped, not the whole repository: SKILL.md at the root, the book_to_skill/ package and scripts/ — 19 files in total; everything else in the repo is its own documentation, tests, evals and site/packaging infrastructure and stays out. Repository: https://github.com/virgiliojr94/book-to-skill (the skill is the mapped perimeter at the pinned commit: SKILL.md at the repository root plus book_to_skill/ and scripts/ — 19 files in total, listed in step 1 under their installed names (which are their repository-relative names); the rest of the repository is not part of the skill) Pinned version: install the code at commit a24e62f4d8caf2ee433f6e48d9073800f56f906e exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/virgiliojr94/book-to-skill into a temporary folder, check out commit a24e62f4d8caf2ee433f6e48d9073800f56f906e, and copy the ./ folder from that checkout and from that checkout take exactly these mapped files: SKILL.md from the root, every file under book_to_skill/, and scripts/extract.py — do not copy docs/, tests/, evals/ or any other repository file. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-22: [FINGERPRINTS] - SKILL.md: SHA-256 6a3d7ce3d3689020cd4089f8cad9c0a9f8237abfaf896ff20677031dac4592b1 - book_to_skill/__init__.py: SHA-256 9d18b99b2c417557244634073c6bb3fa52dcc58aad8fb1fc5f658599f0c3ab77 - book_to_skill/__main__.py: SHA-256 2207592e7bad6433302d258a4cfc9384a406373f59e49392a56d8389d08803ea - book_to_skill/cli.py: SHA-256 2fe257d1c089ca5ecc68c6dde377087766b83d3e9c5393810c6bb5401dc97a91 - book_to_skill/config.py: SHA-256 e92a487f632db762f4dafff3188f6441809b4cba3bb88b0c6821a14e6a21ff2f - book_to_skill/dependencies.py: SHA-256 42ada779fd8d8eed961790f4d49cadd1c824ec7a9cb7fbfd36f02a26c9b12988 - book_to_skill/exceptions.py: SHA-256 bf9417993e577cea5130bc34f4241cafccd1dda1f2f1ac22a9e8a198cc9dedc8 - book_to_skill/parsers/__init__.py: SHA-256 a8f740712820f8729cd3b6adca799d17a23e0b80205a29e511791aafc97723f0 - book_to_skill/parsers/calibre.py: SHA-256 ed2589e86408cb5e82f795a20ce3cf8e32a9f8307fd3e838ba3a7c466f200d42 - book_to_skill/parsers/docx.py: SHA-256 3cf1e87d6d73ffd99916503656783fc70ef2c5e1f3e4d3ab55223c2db22e19eb - book_to_skill/parsers/epub.py: SHA-256 4cb0b38868fa13550e8a189d916889a097990afa59566a565e286c88ec3291cb - book_to_skill/parsers/html.py: SHA-256 dec7b3b84dc514469c050fa71bafcc8109956d2f296198e05d1bb20fce1bc2e6 - book_to_skill/parsers/pdf.py: SHA-256 dd1793ffef319dfa9041627ab34805a799239ddef8503fdcfb071db5816bfcc8 - book_to_skill/parsers/rtf.py: SHA-256 84ebfeaee217fb96c92ef008b57059d139aecd12311e0d317f314422e06b89de - book_to_skill/parsers/text.py: SHA-256 5312d325b3f21c5977c1d7a276a2fda469d4ae81241636ede6e351c7eaacca80 - book_to_skill/pdf_inspector_integration.py: SHA-256 e1f6790be272068a9d82ab2b2fe78cd1510dba12748bf9f9931514b759c6d1c2 - book_to_skill/sanitize.py: SHA-256 d334782ca6aa168e1927d6334f0e6888322f482984e4ab128362e73fd66cce26 - book_to_skill/utils.py: SHA-256 f6ad43fdbd399e579e7ff13c92bb81b2c99ed655413e55ff755734d0b13cbe5e - scripts/extract.py: SHA-256 f0641dc486c0d8323ebdde449dc80b49d1cd94a4aa2320058a2a13a2b48582ac If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md in full, and tell me in plain words: what the skill does when used, what it connects to, whether it reads or sends any personal information or credentials, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - Mapped perimeter, exactly 19 files installed under their original names: SKILL.md (root), book_to_skill/ with 17 Python files (the package: cli, config, dependencies, exceptions, sanitize, utils, pdf_inspector_integration and the parsers/ subpackage for PDF, EPUB, DOCX, HTML, RTF, text and Calibre formats), and scripts/extract.py. The rest of the repository — docs/, tests/, evals/, tools/, .github/, READMEs, mkdocs.yml, pyproject.toml and the other root files — is repository infrastructure and is not copied. - Nothing is sent anywhere and nothing is run during installation itself; the only download at install time is these 19 files from GitHub at the pinned commit. The Python package and extract.py are copied as-is, never executed. - At use time the skill runs locally on documents I provide: its extractor reads them on my machine, it may ask before installing a missing optional extractor (--install-missing ask — it never installs one silently), and it presents a token-cost estimate before generating the skill files. The optional GitHub publication of a generated skill (Step 11) happens only if I ask for it, with the gh CLI and my own authentication. - The generated output is a new skill written to my agent's skills location (the skill prefers the cross-agent ~/.agents/skills/ root, with a Claude Code symlink step it describes); nothing about my books is sent to a third-party service by the instructions I read. - Do not convert, extract or analyse any book or document, do not run extract.py, and do not use the skill in any way during this install: using it is a separate step that only starts when I ask for it later. 3. When I say "yes, install", install it as a skill: copy the 19 mapped files into the agent's skills directory, in a folder named exactly "book-to-skill", keeping their original relative layout (SKILL.md at its root, book_to_skill/ and scripts/ beside it). 4. Verify with no network calls: frontmatter valid (name: book-to-skill), the 19 files in place with matching fingerprints. Do not run anything from the folder and do not use the skill to verify. 5. Report what was installed, where, and what I still need to do myself (for example: that the install perimeter is only these 19 mapped files (not the whole repository), that using it runs its Python extractor locally on my documents, that a missing optional extractor is installed only after I am asked, that it shows a token-cost estimate before generating, and that MOBI/AZW need Calibre). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.