# Books, courses and the paid team pack The hooks in this repository are MIT-licensed and work without any of this. The write-ups behind them are long-form, so they live here rather than in the README. Moved out of the README on 2026-08-31 to keep that file a tool document — see [the note at the bottom](#why-this-is-not-in-the-readme). ## Where these hooks came from Every hook here exists because something broke first. The incident records behind them — what failed, what the logs actually looked like, and what finally stopped it — are written up at length in these: **In Japanese** — these two are the ones the hooks here were actually written against, and they are the deepest: - [事故防止の全記録](https://zenn.dev/yurukusa/books/6076c23b1cb18b) — 97 chapters of incidents, each traced to the setting or hook that stops it. The introduction, the symptom→chapter lookup table you'd reach for mid-incident, Chapters 1-3 and Chapter 100 are free to read - [トークン費用の実測](https://zenn.dev/yurukusa/books/token-savings-guide) (¥2,500) — where the tokens actually go, measured across 800+ hours rather than reasoned about. 35 chapters; the introduction, the symptom→chapter cost table, and Chapter 1 are free to read - [副の作業者の沈黙の失敗](https://zenn.dev/yurukusa/books/sub-agent-observability) (¥1,500) — eight incidents where a sub-agent reported success and had run nothing, sorted into four failure shapes: fabricated dispatch, silent stop, no observability, scope drift - [MCP プラグインの隠れたコストと信頼性](https://zenn.dev/yurukusa/books/mcp-plugin-reliability) (¥800) — five vulnerability classes in third-party MCP plugins and what a user (not the author) can do about each - [AGENTS.md × Claude Code を両立する5つの方法](https://zenn.dev/yurukusa/books/agents-md-interop) (¥1,500) — the Japanese edition of the AGENTS.md Interop Handbook below - [チーム/企業導入 安全パック](https://zenn.dev/yurukusa/books/cc-team-safety-pack) (¥3,000) — the approval write-up, the policy, the permission design and the monthly review, for the person who has to get Claude Code past their own organisation **In English:** - [Claude Code Safety Mastery](https://leanpub.com/claude-code-safety-mastery) (from $9.99, 66 pages) — the defensive hooks in this repository, grouped from the five to install first through Git protection and credential guards, and eight dated incidents where the guard itself failed silently - [Claude Code Migration Playbook](https://leanpub.com/claude-code-migration-playbook) (from $11.99, 251 pages) — stay, switch, or build your own stack: five measurable triggers, a 30-day cost projection for each path, a decision tree that returns one recommendation, and a 48-hour rollback if it was wrong - [Cut Your Claude Code Token Usage in Half](https://leanpub.com/claude-code-token-savings) (from $9.99, 89 pages) — where the tokens actually go, measured across 800+ hours rather than reasoned about: overnight cost spikes, sub-agents, thinking tokens, and context-window bloat - [Claude Code AGENTS.md Interop Handbook](https://leanpub.com/claude-code-agents-md-interop) (from $9.99, 49 pages) — which file each of nine tools reads, six ways to keep them in sync, and how to check what your own setup actually loads rather than trusting a closed issue - [CLAUDE.md Under Test](https://leanpub.com/claude-md-under-test) (from $9.99, 86 pages) — thirty-nine trials on whether a rule written in `CLAUDE.md` is actually obeyed, what a hook adds once it is, and the exit code that decides whether your guard fails open or closed. The rule was obeyed in all twenty-two trials where it was written, which is the opposite of what I had been telling people — the correction is posted free in [Discussion #59](https://github.com/yurukusa/cc-safe-setup/discussions/59). Every trial's data and the harness are in the appendices **Start with the free one.** [The Claude Code Safety Field Manual](https://leanpub.com/claude-code-safety-field-manual) (124 pages, minimum price zero) is this repository's documentation laid out as a path: the pre-flight checklist, what each guard actually refuses, how to make one fire on purpose so you can watch it work, and what to read in the log afterwards. Everything in it is also in this repository. Nothing here is gated behind it. The five above and two more (the Field Manual, and *Claude Code Safety*, 68 pages) are also sold together as [Claude Code Operator's Library: Complete Edition](https://leanpub.com/b/cc-operators-complete) (seven titles, from $39) — cheaper than the five paid titles bought separately. The older four-title bundle, [The Claude Code Operator's Library](https://leanpub.com/b/cc-operators-library) (from $29), is still on sale for anyone who only wants those four. **Every paid title has a free sample you can read before deciding.** **More than one person.** Each of the five is also sold at a team discount — up to 3, 5, 10, 15 or 25 members in a single purchase, one order and one receipt, with no quote and no sales call. The per-member price falls as the count rises: for a $9.99 book, three members cost $24 rather than $29.97, and twenty-five cost $149 rather than $249.75. The $11.99 Migration Playbook is $29 for three and $179 for twenty-five. Same book, same files; only the number of people covered differs. The bundle and the course have no multiple-copy packages, so a team that wants the whole library buys each book's team discount separately. The 50-point checklist in `audit/` also exists as a scored course, if you would rather work through it as lessons with quizzes than run the checklist yourself: [The Claude Code Safety Audit](https://leanpub.com/c/claude-code-safety-audit) (from $49) — six sections, the one-command test that hands a hook the operation it should refuse and reads the exit code, and the eight incidents where the guard failed silently. It uses the same four scripts in `audit/`, which stay free and MIT-licensed here. Two of them are on Gumroad as well, if you prefer that store: [Migration Playbook](https://yurukusa.gumroad.com/l/claude-code-migration-playbook) ($19) and [the token book](https://yurukusa.gumroad.com/l/azrdt) (¥2,500). All of them are optional, and every hook in this repository works without them. The reason they are listed at all is that the Japanese editions do not surface in search or in Zenn's own topic listings for this account (measured across 19 topics on 2026-08-09: zero appearances), so this page is one of the few places they can be found from. ## The paid team pack One paid thing exists, and it needs no inquiry either: the [Team Safety Rollout Pack](https://yurukusa.booth.pm/items/8230188) (¥3,000, one-off) bundles the shared policy template, the CI gate with defaults already chosen, and a written walkthrough of real reported incidents. ## Written audits Three asynchronous written audits exist ($29 and $219): no call, no meeting, and nothing is ever run in your environment. Each publishes its full deliverable before you buy, run against my own setup. Scope, prices and the samples are in [SERVICES.md](../SERVICES.md). Corporate audits, training, rollout consulting and monthly retainers are **not** offered. ## Why this is not in the README On 2026-07-02 this repository's README was cut from 778 lines to 74 and every paid link removed, as part of the compliance work promised to GitHub after an account-level anti-spam flag (the one that made every page under `yurukusa.github.io` return 404). GitHub's Acceptable Use Policy §10 rules out repositories whose primary purpose is promotion. Between July and August the README grew back to 302 lines and 13 paid links, one justified pull request at a time, without anyone deciding to take that risk again. This page is where that content lives now.