Centaur Review transforms the code review process into a collaborative dialogue between you and Claude, leveraging a 'centaur' approach where the human leads and the AI assists. Instead of manually formatting every comment, you narrate your thoughts as you read the code while Claude tracks observations, maps them to specific line numbers, and runs five parallel background agents to detect bugs, architectural flaws, and requirement gaps. The skill includes a unique 'Red Teamer' phase that challenges your suggestions to ensure technical accuracy, ultimately helping you generate a polished, friendly, and actionable GitHub review with professional formatting and severity badges.
