An agentic code review loop where every developer fires their own AI review skill at a pull request, the author's agent answers, and a scorecard rates each.