ln-623-code-principles-auditor_skill

Overview

This skill audits code principles (DRY, KISS, YAGNI), error handling, and DI/init patterns to produce structured findings and a compliance score. It runs targeted pattern-based scans, maps each finding to refactoring recommendations, and outputs a single report file with an extended JSON block for cross-domain DRY analysis. The worker is domain-aware and returns severity, location, effort, pattern_id, and pattern_signature for each finding.

How this skill works

The auditor parses the coordinator-provided contextStore to determine scan_path, tech_stack, principles, and output_dir, then loads language-specific detection patterns. It runs Grep/Glob detections constrained by exclusions, matches findings to refactoring patterns via the decision tree, and tags each result with severity, effort, location, and pattern_signature. Finally it computes a penalty-based score, writes the full markdown report (including a <!-- FINDINGS-EXTENDED --> JSON block for DRY findings) in a single write, and returns a concise summary to the coordinator.

When to use it

• During automated codebase quality gates before major releases
• As part of continuous architecture reviews or technical debt sprints
• When onboarding a new domain or verifying cross-domain duplication
• Before refactoring initiatives to prioritize high-impact duplication
• To validate error handling and DI patterns across services

Best practices

• Run domain-aware scans to limit noise and surface domain-specific issues
• Ensure detection_patterns and refactoring_decision_tree references are up-to-date for accurate matching
• Exclude generated, vendor, and migrations directories to avoid false positives
• Treat pattern_signature as the canonical key for cross-domain DRY correlation
• Use effort estimates (S/M/L) to prioritize fixes alongside severity

Example use cases

• Detect identical authentication logic duplicated across microservices and recommend extraction into a shared module
• Find async route handlers missing try/catch and flag critical user-facing endpoints as high severity
• Identify unnecessary factories or single-implementation abstractions and recommend simplification
• Surface repeated validation logic (email/password) with a pattern_signature for cross-repo consolidation
• Verify presence of a centralized ErrorHandler and bootstrap/DI initialization patterns