Code Quality Auditor

What it does

The Code Quality Auditor skill analyzes backend code changes to identify potential issues and ensure adherence to coding standards, security protocols, and testing requirements. It provides a structured review process, generating reports with conclusions (Pass or Reject), problem severity levels (blocker, warning, suggest), and detailed evidence of verification steps taken. The tool helps improve code reliability, maintainability, and overall quality by systematically identifying bugs and style violations across various file types including documentation, configuration files, and agent/skill definitions.

When to use it

• Before merging code changes into a main branch.
• During code review processes for new features or bug fixes.
• To ensure compliance with coding standards and security guidelines.
• When verifying the quality of documentation, scripts, configuration files, and skill definitions.
• For release preparation to catch potential issues before deployment.

Key capabilities

• Contextual Review: Reads git diffs, change file lists, Todo acceptance points, and existing verification results to understand the scope of changes.
• Minimum Verification Enforcement: Determines minimum validation requirements based on a "Verification Matrix" and enforces necessary commands (e.g., pnpm lint, pnpm typecheck).
• Structured Assessment: Uses a "Review Checklist" to cover correctness, security, responsibility boundaries, test coverage, and documentation consistency.
• Evidence Generation & Tracking: Generates evidence records and tracks changes across multiple review rounds.
• Problem Classification: Categorizes issues as blocker (critical), warning (potential risk), or suggest (improvement).

Example prompts

• "Analyze the code changes in this git diff for quality issues."
• "Review these files [list of file paths] and provide a Code Quality Auditor report."
• “Perform a review gate on my latest commit, ensuring all minimum verification requirements are met.”

Tips & gotchas

• Prerequisites: The AI agent needs access to the codebase's git repository.
• Minimum Verification is Key: The skill will reject code that doesn’t meet the defined "Verification Matrix" even if no immediate errors are found.
• Evidence is Essential: A “Pass” conclusion can only be given when all blocker issues are resolved and minimum verification requirements are met, with clear evidence provided.