parallel-code-review

Parallel Code Review

This skill provides guidance for launching multiple specialized code review agents in parallel for comprehensive, efficient analysis from different perspectives.

Purpose

Parallel code reviews maximize efficiency and coverage by running multiple specialized reviewers simultaneously. Instead of sequential reviews that take time proportional to the number of reviewers, parallel execution completes in the time of the slowest reviewer while providing comprehensive feedback from all perspectives.

When to Use This Skill

Use this skill when:

• Performing comprehensive code review before merging
• Need multiple specialized perspectives (security, architecture, performance)
• Want faster review by parallelizing analysis
• Reviewing large changesets that benefit from division of labor
• Implementing continuous review practices

Benefits of Parallel Reviews

Speed: 2+ specialized reviews complete in the time of 1 Depth: Each agent focuses on specific expertise area Comprehensive Coverage: Security + Architecture + Performance simultaneously

Core Workflow

Phase 1: Prepare for Review

1. Check Decision Log (Prevent Redundancy)

# Search memory for previous code review decisions
mcp__memory__search_nodes query:"code_review_decision"

# Read decision log file
cat code_review_decisions.md

Decision log format:

# Code Review Decisions

## 2025-01-15: Result Pattern Required

**Decision**: All service objects must return Result objects
**Rationale**: Explicit success/failure handling improves error management
**Status**: Accepted standard pattern

2. Get Code Changes

# Get diff for review
git diff main...HEAD

# Or specific branch
git diff main...feature-branch

Phase 2: Launch Parallel Reviewers

Use Task tool to launch multiple agents concurrently:

Example: Launch 2 reviewers in parallel by sending a SINGLE message with MULTIPLE Task tool calls:

Task({
  subagent_type: "cybersecurity-expert",
  description: "Security review of changes",
  prompt: "Review git diff for security vulnerabilities..."
})

Task({
  subagent_type: "rails-backend-expert",
  description: "Architecture review of changes",
  prompt: "Review git diff for code quality..."
})