Auto Code Reviewer is a skill that acts as an automated quality gate for code produced or modified by an AI agent. Every piece of code passes through three sequential review passes before it is returned to the user, with issues identified and fixed automatically rather than merely flagged.
The first pass focuses on error handling and robustness. It wraps API calls and file I/O in try/catch blocks with descriptive error states, adds null checks, validates inputs, handles empty datasets, and eliminates unhandled promise rejections or silent failures.
The second pass addresses structure and cleanliness. Functions exceeding 30 lines are refactored automatically. Repeated logic is extracted into shared utilities, lookup data is pre-fetched to remove N+1 query patterns, and dead code along with unused imports is stripped out.
The third pass handles naming and simplification. Vague variable names such as data, temp, or result are replaced with intent-revealing names. Complex conditionals are rewritten using well-named booleans, nested logic is simplified, and magic numbers are extracted into named constants.
This skill integrates directly into an AI agent workflow, applying all three passes consistently on every code generation or modification without requiring manual prompting. Unlike a suggestion-only approach, it auto-fixes every issue it identifies. It is a good fit when consistent, automated enforcement of code quality standards across agent-generated output is a priority.