/cs:book-to-skill <path|folder|glob>... [skill-name] — convert a book, documentation folder, or source collection into a structured agent skill (core frameworks
复制下面这句话，粘贴给 Claude Code、Codex、Cursor 等 AI 编程工具，它会读取安装说明并在你确认后完成安装。
请阅读 https://ai.atlankj.com/install/asset/gh-claude-skills-f989f4032f96 ，按照其中的说明把「cs-book-to-skill」安装到你（当前 AI 工具）中。执行前先告诉我将运行的命令和写入的位置，等我确认。
查看 AI 将读取的安装说明正在读取 GitHub 原文…
内容来自 GitHub 原始文件，由原作者维护。在 GitHub 查看
Command: /cs:book-to-skill <path|folder|glob>... [skill-name-slug]
Runs the converter end to end: extract → analyze → chapter files → supporting files → master
SKILL.md → validate. Add "analyze only" to stop after the extraction report.
The command refuses, with a reason, when:
| Gate | Refusal |
|---|---|
| No path given | Prints usage. This tool converts files on disk — not titles from memory, not URLs. |
| No supported file resolves | Names what was searched and the supported extensions. |
| Source is smaller than ~3× the compiled skill | Says converting is not worth it and recommends handing the agent the document. |
| Cost estimate not approved | Waits. Generation is the expensive step and the user approves it with numbers in front of them. |
| Validation errors after generation | Blocks. Dead chapter links and dangling topic references break navigation silently. |
Asked one at a time, each with a recommended answer.
Recommended: convert when it is > 3× the compiled skill's size and you will return to it.
One-shot reads are cheaper unconverted. token_budget_estimator.py prints the verdict.
Recommended: reference, unless you intend to internalize the author's reasoning. Study depth roughly doubles generation cost and only earns it with real worked examples.
Recommended: technical only when tables, code, or formulas carry meaning. Docling costs ~1.5s/page and buys nothing on a prose book.
Recommended: name three real questions before generating. They decide what belongs in Core Frameworks and what the topic index must resolve.
Recommended: assume not. Keep it local unless the source is public-domain, openly licensed, your organisation's own documentation, or you have written permission.
Recommended: check for a compiled skill on the same subject first. Folding new sources into one skill beats two skills that half-cover a topic and give the agent no way to choose.
SKILL_ROOT=engineering/book-to-skill/skills/book-to-skill
# 0. environment (optional — reports extractors, installs nothing)
python3 "$SKILL_ROOT/scripts/extract_document.py" --check
# 1. extract
python3 "$SKILL_ROOT/scripts/extract_document.py" <paths> --mode text|technical
# 2. worth-it verdict, before spending a generation pass
python3 "$SKILL_ROOT/scripts/token_budget_estimator.py" --full-text "$WORKDIR/full_text.txt"
# 3. generate (agent work: chapters, glossary, patterns, cheatsheet, SKILL.md)
# 4. gate
python3 "$SKILL_ROOT/scripts/book_skill_validator.py" "$SKILLS_HOME/<slug>"
python3 "$SKILL_ROOT/scripts/token_budget_estimator.py" --skill-dir "$SKILLS_HOME/<slug>"
✅ <slug> — <Title> by <Author> <N> chapters
SKILL.md ~<N> tokens (resident) · chapters ~<N> each (on demand)
validator: <N> error(s), <N> warning(s)
next: /cs:book-to-plugin to package it for this repo
/cs:book-to-plugin — wrap a compiled skill as a claude-skills plugin/cs:write-a-skill — author a skill from your own expertise instead of a document