code-review

name	code-review
description	Structured code review with severity-based findings
arguments	[{"name":"target","description":"File path or glob pattern to review","required":true},{"name":"focus","description":"Review focus area (security|performance|readability|all)","default":"all"}]

Code Review Skill

Process

1. Read the target file(s): {{target}}
2. Check against project standards defined in root CLAUDE.md
3. Analyze for: type safety, error handling, Hono API patterns, Drizzle ORM usage, security
4. Focus area: {{focus}}

Output Format

For each finding, output:

[SEVERITY] Title

• File: path/to/file.ts:lineNumber
• Category: security | performance | readability | maintainability | correctness
• Finding: Description of the issue
• Suggestion: Concrete fix with code example
• Reference: Link to relevant docs or ADR if applicable

Severity levels: CRITICAL | HIGH | MEDIUM | LOW | GOOD (positive callout)

Checklist

• No any types without justification
• All async operations have error handling
• Hono middleware used correctly (context flow, error propagation)
• Drizzle ORM queries use proper typing (no raw SQL without justification)
• Zod schemas validate all external input
• No hardcoded strings (use constants or environment variables)
• Functions ≤ 50 lines
• Files ≤ 300 lines
• No circular imports
• SSRF protection on external URL fetching (safe-fetch.ts)
• Copyright constraint respected (no body_original/body_translated in public responses)