Multi-AI Code Review

Overview

multi-ai-code-review provides comprehensive code review using multiple AI models as specialized agents, each analyzing code from a different perspective. Based on 2024-2025 best practices for AI-assisted code review.

Purpose: Multi-perspective code quality assessment using AI ensemble with human oversight

Pattern: Task-based (5 independent review dimensions + orchestration)

Key Principles (validated by tri-AI research):

1. Multi-Agent Architecture - Specialized agents for each review dimension
2. LLM-as-Judge Consensus - Flag issues only when 2+ models agree
3. Progressive Severity - Critical → High → Medium → Low prioritization
4. Human-in-Loop - AI suggests, human decides
5. Quality Gates - Block merges for critical unresolved issues
6. Actionable Feedback - Every comment has What/Where/Why/How