Code Review

Part of Agent Skills™ by googleadsagent.ai™

Description

Code Review enforces a structured pre-merge quality gate with a checklist-driven evaluation, severity-classified findings, and mandatory resolution tracking. The agent reviews every diff against a configurable set of quality dimensions before approving changes, ensuring consistent standards regardless of reviewer fatigue or time pressure.

Unlike ad-hoc review comments, this skill produces a standardized review document with findings categorized by severity: Critical (must fix before merge), High (should fix before merge), Medium (fix in follow-up), and Low (optional improvement). Each finding includes the file, line range, category, description, and a concrete suggested fix. The review document becomes part of the permanent record.

The pre-review checklist catches common oversights before deep analysis begins: missing tests, uncommitted files, linter errors, type errors, and documentation gaps. Only after the checklist passes does the agent proceed to semantic review of logic, architecture, security, and performance.

Use When

• A pull request or diff is ready for review
• The user asks for feedback on code changes
• Before merging any branch into main
• After a subagent completes a task (Stage 2 review)
• Code has been refactored and needs validation