Requesting Code Review

Automate code reviews by dispatching an AI subagent to analyze your code changes against a plan or requirements. The subagent evaluates quality, architecture, and testing, then produces a structured report with issues by severity and actionable recommendations, enabling faster iteration without waiting for human review.

Request code review after completing tasks in subagent-driven development, after finishing major features, before merging to main, or when stuck or planning refactoring. Early checkpoints catch issues before they compound, reducing rework and accelerating your development cycle.

Provide the BASE_SHA and HEAD_SHA from git to identify your changes, a description of what was implemented, and the plan or requirements document your changes should satisfy. The subagent uses this context to evaluate your code against your stated goals and detect misalignments.

Yes, dispatch the code-reviewer subagent as part of your development workflow after task completion or before merging. By capturing git SHAs and relevant plan documentation, you can automate review checkpoints at any stage and receive structured feedback without manual intervention.

The subagent produces a detailed reviewer report covering code strengths, issues organized by severity, architectural concerns, testing gaps, compliance with requirements, and an overall assessment. Recommendations are actionable, allowing you to address feedback immediately and iterate faster.

Basic git familiarity is helpful—you'll retrieve BASE_SHA and HEAD_SHA using standard commands like `git rev-parse HEAD` to identify your commit range. The skill handles dispatching the review; git knowledge focuses on capturing the change boundaries accurately.