requesting-code-review

To automate AI code review before a pull request merge, you provide a summary of changes, the plan or requirements, and the git commit range. The skill dispatches a code-reviewer subagent to analyze the changes and highlight critical issues, important issues, and minor improvements.

Starting an AI-assisted code review requires three inputs: a summary of what was implemented, the plan or requirements reference, and the git commit range (base SHA and head SHA). This allows the code-reviewer subagent to evaluate alignment and code quality.

You should request a formal code review after completing each subagent-driven task, after finishing major features, or before merging to main. Ad-hoc reviews can also be triggered whenever you need automated feedback on recent code changes.

An AI code reviewer evaluates changes against a project plan by comparing the provided implementation summary and git diff with the original requirements. It identifies misalignments and generates actionable feedback categorized by critical, important, and minor issues.

Yes, you can perform an ad-hoc code review on recent git changes without a full merge by providing the necessary git range and implementation summary. The skill supports triggering reviews flexibly whenever you need to assess code quality outside the standard task workflow.