A Structured, 5-Level Deep Code Analysis

Ditch the generic "looks good to me" feedback. This skill implements a rigorous, multi-stage code review protocol designed to find what simple prompts miss. By forcing the agent through mandatory checkpoints, it ensures that security, performance, and test coverage are never secondary to getting the code merged.

Why use this skill?

Standard AI prompts often suffer from "reviewer fatigue" or hallucinate that code is safe just because it's readable. This skill solves that by enforcing a disciplined 5-level hierarchy:

• High-Level Architecture: Validates patterns and design consistency first.
• Security First: Mandatory deep-dives into auth, input validation, and data exposure.
• Verification: Analysis of test coverage and identification of missing edge cases.
• Performance Analytics: Identifies scalability bottlenecks and resource leaks.
• Structured Output: Delivers a standardized markdown report with a clear final recommendation (Approve/Request Changes).

Compatibility

While optimized for the Grok Build CLI/TUI, this skill is compatible with any agent that can read project files, including Claude Code and Cursor. It is language-agnostic, supporting TypeScript, Python, JavaScript, and more.

Output Format

The result is a professional-grade audit report featuring a severity-sorted findings list, a testing assessment, and an actionable list of required changes to get the PR to a mergeable state.