requesting-code-review

To automate code review, route your code changes to a dedicated subagent that evaluates task results against the plan. This process catches issues early and prevents cascading defects before merging to main.

A structured code review requires WHAT_WAS_IMPLEMENTED, PLAN_OR_REQUIREMENTS, BASE_SHA, HEAD_SHA, and DESCRIPTION. Providing this context ensures the reviewer subagent produces a thorough, prioritized assessment of your changes.

You should run an automated code review after completing tasks, after major feature completions, or before merging to main. This timing ensures quality assurance checks catch issues early in the development cycle.

A subagent code review prevents regressions by evaluating your implemented changes against the original plan and dispatching actionable feedback with a clear severity taxonomy and remediation steps.

The best way to dispatch a focused code review is to route your base and head SHAs along with a summary of changes to a dedicated code-reviewer subagent, triggering a focused evaluation of your code.