review-booklisted

# Skill: review-book **What it does.** Produces the whole-book report: literary review, commercial assessment, and technical AI-use analysis. The wide-angle counterpart to `analyze-chapter`'s microscope. **Triggers** - "review the book" - "/review-book" - "run the full report" - "review Act {N}" - Equivalent phrasing in the project's output language **Input.** Scope: whole book (default) or a single Act. **Output.** `.project/reports/literary/YYYY-MM-DD-review-{scope}.md`, following `.project/templates/book-review.md`. > **Write the report in the project's output language.** --- ## Core principle > **Honesty over kindness, precision over both.** A report that only praises is useless. One that only criticizes is demoralizing and equally useless. The target: what works *and why*, what fails *and how to fix it*, with evidence for every claim. Three audiences read this report: the author deciding what to revise, the author deciding whether to submit, and — potentially — the material behind an honest AI-use declaration. All three need evidence, not adjectives. --- ## Execution protocol ### Step 1 — Load context 1. `.project/config/project.yaml` — output language, genre, targets, `paths.layout` 2. **The entire manuscript in scope** — read fully, in order. No skimming; structural judgments require having actually felt the pacing. Resolve the chapter list and its ordering via `.project/templates/layout.md`; an Act scope resolves from the chapter id prefix 3. `.project/k