When you need multiple people reviewing code from different angles (security, performance, architecture), this coordinates the whole process and prevents chaos. It gives you clear rules for which dimensions to assign based on what changed, handles the messy job of deduplicating findings when two reviewers flag the same line for different reasons, and calibrates severity ratings so "critical" means the same thing to everyone. The consolidated report template structures everything into a single actionable document with finding counts by dimension. Saves you from manually merging conflicting feedback and arguing about whether that SQL injection is "high" or "critical."
npx skills add https://github.com/wshobson/agents --skill multi-reviewer-patterns