requesting-code-review

Automate a code review by dispatching a reviewer subagent to evaluate completed work against your plan. You provide a task description, requirements, and Git base and HEAD SHAs to compare, and the system returns a structured prompt with calibration guidance.

Run an automated code review after each completed task in subagent-driven development, after major features, and before merging to main. This applies the core principle of reviewing early and often to catch issues before they compound.

Generating a structured code review prompt requires a task description, a plan or requirements, and the base and HEAD Git SHAs to compare. These inputs allow the reviewer subagent to focus strictly on the work product.

Automated code review evaluates only the final work product, not the author's session history. The reviewer subagent focuses strictly on the completed changes defined by the provided Git ranges and requirements.

You can integrate a formal code review stage into standard workflows by applying it after each subagent task, major features, and pre-merge. This ensures continuous quality assurance throughout the development lifecycle.

The best way to catch issues early is to dispatch a general-purpose reviewer subagent after each task. It evaluates your changes against explicit plans using a structured prompt, providing precise feedback before you proceed.