After
Skill diguike/book-claude-skill/examples/ch17-bloat-refactor/after
Claude Code Skill 开发指南
npx -y skills add diguike/book-claude-skill --skill afterAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Code review with security, performance, and style checks. 代码审查:安全、性能、风格。
SKILL.md
2.8 KB, as published. Nobody here has run it
Code Review
Review changed files and report issues by severity.
Workflow
- Run
git diff --name-only HEAD~1to collect changed files - For each file, read the content and diff
- Apply rules in order (stop-on-error for security):
rules/security.md— MUST pass, any error blocks deploymentrules/performance.md— warnings and errorsrules/style.md— warnings only
- Generate structured report (see output format below)
- Log metrics: run
scripts/append-log.tswith the results
Output Format
# Review: [branch-name]
## Summary
- Files: X | Issues: Y (E errors, W warnings)
## Security (rules/security.md)
| File | Line | Issue |
|------|------|-------|
## Performance (rules/performance.md)
| File | Line | Issue |
|------|------|-------|
## Style (rules/style.md)
| File | Line | Issue |
|------|------|-------|
Configuration
Skip patterns: *.test.ts, *.spec.ts, *.d.ts, *.min.js
Max file size: 10000 lines (skip larger files)
Constraints
- Do NOT auto-fix code — only report issues
- Do NOT send notifications — that is the CI pipeline's job
- Do NOT generate trend reports inline — use
scripts/trend-report.tsseparately - Keep the report under 200 lines; link to files instead of quoting large blocks
Rules
Rules are loaded from rules/ directory:
rules/security.md— injection, XSS, secrets, auth bypassrules/performance.md— N+1, bundle size, memory leaks, framework-specificrules/style.md— naming, complexity, dead code, formatting
References
Loaded conditionally based on file types in the changeset:
references/react.md— when .tsx/.jsx files are presentreferences/vue.md— when .vue files are presentreferences/typescript-strict.md— when tsconfig strict mode is enabledreferences/api-design.md— when files in src/api/ or src/routes/ are changed
Scripts
scripts/collect-metrics.ts— extract metrics from review resultsscripts/append-log.ts— append metrics to data/review-log.jsonlscripts/trend-report.ts— generate trend report (run separately)scripts/score.ts— calculate review score from issues
What Was Removed (vs the bloated version)
The original 300+ line SKILL.md included:
- Inline security rules (→ moved to
rules/security.md) - Inline performance rules (→ moved to
rules/performance.md) - Inline style rules (→ moved to
rules/style.md) - Auto-fix logic (→ removed; separate skill responsibility)
- Deploy notification (→ removed; CI pipeline responsibility)
- Trend report generation (→ moved to
scripts/trend-report.ts) - Detailed code examples for every rule (→ moved to
references/)
Result: 300+ lines → ~150 lines. Each concern in its own file.