Multi-Model Code Review

Overview

This skill provides guidance for running code reviews across multiple LLMs in parallel and consolidating findings into a unified, high-confidence report. The core principle is consensus-based triage — issues flagged by multiple independent models carry more weight than those flagged by only one.

When to Use This Skill

• MultiReviewer is performing consensus consolidation
• Evaluating conflicting findings from different models
• Triaging false positives vs. real issues
• Deciding review depth (single-model vs. multi-model)

When NOT to Use This Skill

• Standard single-model review is sufficient
• Simple or low-risk code changes
• Quick fixes or configuration updates