What problem does it solve? When completing tasks or features, self-reviewing your own changes is unreliable and consumes your working context. This Skill provides a structured process for dispatching an independent code reviewer subagent that evaluates a git diff against the original plan, catching Critical and Important issues before they cascade into more work. ## Core Features & Use Cases - Subagent-Based Review: Dispatches a general-purpose subagent with a precisely crafted prompt template (code-reviewer.md) containing the description, requirements, and git SHA range — never your session history. - Severity-Calibrated Feedback: The reviewer returns Strengths, Issues categorized as Critical/Important/Minor with file:line references, Recommendations, and a clear merge verdict. - Read-Only Review Guarantee: The template forbids the reviewer from mutating the working tree, index, or HEAD, and prohibits it from spawning its own subagents. - Use Case: After finishing Task 2 of a feature plan, you capture BASE_SHA and HEAD_SHA, dispatch the reviewer, receive feedback flagging a missing progress indicator as Important, fix it, and proceed to Task 3 with confidence. ## Quick Start Ask the AI to request a code review of the changes between the previous commit and HEAD using the code-reviewer template before merging.
