@endgame-build/claude-workflows/.claude/commands/eg/review.md

Claude Code workflow system with commands, agents, and templates for AI-native development

---
description: Perform thorough code review with actionable feedback
argument-hint: <PR URL or branch name>
---

# /eg:review

You are tasked with performing a thorough code review with actionable feedback. This is a two-phase process that analyzes changes for issues and provides constructive feedback.

Input: `$ARGUMENTS`

## Step 1: Parse Review Target

The entire argument is the review target:
- PR URL (e.g., https://github.com/org/repo/pull/123)
- Branch name (e.g., feature/user-authentication)

## Step 2: Gather Changes

Based on target type:

### For PR URL:
```
!gh pr view {pr-url} --json title,body,author,files
!gh pr diff {pr-url} | head -2000
```

### For branch name:
```
!git fetch origin {branch-name} 2>/dev/null || echo "Branch might be local"
!git diff main...{branch-name} --stat
!git diff main...{branch-name} | head -2000
```

## Step 3: Phase 1 - Summarize Changes

First understand what changed:

```
Task(
  description="Summarize changes for review",
  prompt=`
    Summarize the changes for code review.
    
    Use the summary template from .claude/templates/summary.template.md
    
    Focus on:
    - What functionality is being added/changed
    - The implementation approach taken
    - Scope and impact of changes
    - Any architectural decisions made
    
    This summary will provide context for the detailed review.
  `,
  subagent_type="summarizer"
)
```

## Step 4: Phase 2 - Detailed Code Review

Perform thorough code review:

```
Task(
  description="Perform detailed code review",
  prompt=`
    Perform a thorough code review of the changes.
    
    Use the review template from .claude/templates/review.template.md
    
    Check for:
    
    Functionality:
    - Does the code do what it claims?
    - Are edge cases handled?
    - Is error handling appropriate?
    
    Security:
    - Input validation
    - Authentication/authorization
    - SQL injection, XSS risks
    - Sensitive data handling
    
    Performance:
    - Algorithm efficiency
    - Database query optimization
    - Caching opportunities
    - Memory usage
    
    Maintainability:
    - Code clarity and documentation
    - Proper abstraction levels
    - DRY principle adherence
    - Test coverage
    
    Read CLAUDE*.md files for project conventions.
    
    Structure feedback as:
    - Summary: Overall assessment
    - Strengths: What was done well
    - Critical Issues: Must-fix before merge
    - Suggestions: Improvements to consider
    - Nitpicks: Minor style/convention items
    - Questions: Clarifications needed
    
    Provide specific, actionable feedback with code examples.
  `,
  subagent_type="code-reviewer"
)
```

## Step 5: Format Review

Create actionable review output:

```
!echo "# Code Review: {target}
Date: $(date)
Reviewer: Claude

## Summary
{overall-assessment}

## Strengths ✅
{good-practices}

## Critical Issues 🚨
{must-fix-items}

## Suggestions 💡
{improvements}

## Questions ❓
{clarifications}

## Next Steps
{recommendations}
" > review-$(date +%Y%m%d-%H%M%S).md
```

## Review Guidelines

- Focus on substantive issues over style
- Provide actionable feedback with examples
- Balance criticism with recognition
- Consider the PR's stated goals
- Suggest alternatives, don't just criticize
- Be respectful and constructive

The review should help the developer improve the code while maintaining a positive collaborative environment.