requesting-code-review

To run a targeted code review on a specific git diff, provide the base and head SHAs along with your task requirements to dispatch a reviewer subagent. This evaluates only the actual changes against quality standards.

You should request an automated code review before a merge to main, after completing major features, or after subagent-driven development tasks. This early issue detection prevents bugs and requirement drift from compounding.

Severity-based feedback triage for code reviews categorizes issues into Critical, Important, and Minor levels. This guides you on what to fix immediately versus later, culminating in a clear merge readiness verdict.

Yes, you need base and head SHAs to evaluate code quality in this workflow. Providing precise commit ranges ensures the reviewer subagent evaluates the actual diff rather than your full session history.

The best way to catch requirement drift after subagent development is to dispatch a focused code reviewer with a structured task template. This validates your actual diff against the original plan to prevent propagation.