ワンクリックで
code-review
Automated post-generation code review skill. Checks new/modified code against project standards.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Automated post-generation code review skill. Checks new/modified code against project standards.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Performance and security audit for code quality. Checks memory leaks, OOM risks, startup speed, ANR, lag, and security vulnerabilities. Auto-triggered by proactive-correction agent.
Initialize AI coding assistance framework for this project. Scans source code, generates customized rules based on actual project structure, and creates comprehensive references documentation. Use when setting up a new project or re-initializing after major changes.
Bootstrap AI coding assistance into ANY project. Supports 4 AI tools — Claude Code, Qoder, Codex, OpenCode. All tools share EXACTLY the same split-file architecture (rules/skills/agents/hooks/references loaded on-demand). Only the entry file name and config directory name differ.
{{t "code_review.description"}}
{{t "plan_mode.description"}}
代码生成后的自动化审查技能。对新增/修改的代码执行项目规范合规检查。
| name | code_review |
| description | Automated post-generation code review skill. Checks new/modified code against project standards. |
Trigger conditions (any one):
- Completed multi-file code generation
- Completed code modification involving architecture boundaries
- User explicitly requests "review / check"
plan_modeplan execution completedSkip conditions: Single-line changes, pure config changes, user says "no review needed".
| # | Check Item | Check Method |
|---|---|---|
| 1 | Module dependency direction violation | Does lib/ import from app/ or components/? |
| 2 | Forbidden pattern usage | Search patterns in project_rule.md §3 |
| 3 | 'use client' on layout.tsx | Check all layout.tsx for the directive |
| 4 | Server Component using React hooks | Check for useState/useEffect without 'use client' |
| 5 | Missing error.tsx for async routes | Routes with async data fetch must have error.tsx |
| 6 | Hardcoded keys/privacy data | Search for plaintext API keys, tokens |
| 7 | Server/Client boundary violation | Is a Server Component imported into a Client Component? |
| 8 | any type usage | Search for : any or as any |
| 9 | Missing revalidation after mutation | Does each Server Action call revalidatePath/revalidateTag? |
| # | Check Item | Check Method |
|---|---|---|
| 10 | Hardcoded strings/colors/dimensions | Color hex, pixel values in code |
| 11 | Raw <img> tags | Search for <img in tsx files |
| 12 | <head> tag instead of Metadata API | Search for Head imports |
| 13 | Missing loading.tsx for slow routes | Routes with >500ms fetch |
| 14 | Large client bundle from heavy imports | Large libs in Client Components without dynamic import |
| 15 | Missing TypeScript types for props | Components with untyped props |
| # | Check Item | Check Method |
|---|---|---|
| 16 | Component too large | Components with 300+ lines |
| 17 | Missing dynamic imports | Heavy components below the fold |
| 18 | Code reuse | Similar code blocks extracted? |
| 19 | Metadata completeness | Each page exports metadata? |
| 20 | Accessibility | Interactive elements have ARIA attributes? |
## Code Review Report
**Review scope**: [Changed file list]
**Overall result**: Pass / N warnings / N fatal
### Fatal Issues
### Warnings
### Optimization Suggestions
### Verified
Code generation complete
↓
[Step 0] Pre-review correction — delegate proactive-correction agent
↓
[Step 1] Collect changed file list
↓
[Step 2] Check: fatal → warning → suggestion
↓
[Step 3] Output review report
↓
├── Fatal → Notify user + provide fix
└── Pass/Warning → Output report, continue
proactive-correction agentarch-review agentresource-sync agent