requesting-code-review

To automate code review before merging, dispatch a code-reviewer subagent with a bounded git diff range, the implementation summary, and the original plan. This orchestrates a targeted review to catch defects and requirement mismatches before work is considered complete.

Subagent-driven development involves dispatching focused AI agents for specific tasks. You should request a code review after each significant task, after completing major features, and immediately before merging to main to prevent issues from compounding.

For an effective AI code review, provide a bounded git SHA range, what was implemented, the plan or requirements, and a description of the changes. This scoped, evidence-based approach lets the reviewer judge the work without full session history.

Handle issue severities by applying Critical fixes immediately, Important fixes before continuing work, and tracking Minor issues for later. You should push back with technical reasoning when the review feedback is wrong.

Yes, you can use a code review subagent to validate requirements compliance by providing the original plan or requirements alongside the git diff. The reviewer evaluates the work product against these specific requirements to prevent non-compliant changes.