Requesting Code Review

This is a protocol for dispatching code reviewer agents at the right moments in your workflow. It's built around keeping the reviewer's context clean by passing only git SHAs, requirements, and what changed, not your entire session history. The template handles the busywork of framing review requests and sets clear severity levels for feedback. What's smart here is the timing guidance: mandatory after each subagent task and before merges, optional when stuck. The core insight is that reviews catch compounding mistakes early, which matters more in agent driven development where one bad task can poison everything downstream. If you're running multi agent workflows, this gives you a consistent handoff pattern.