What problem does it solve? Manually reviewing Java code for code smells like duplicated logic, god classes, deep nesting, and unused methods is slow and inconsistent. This Skill performs a static heuristic audit of Java source files and produces a fixed-shape report of findings with exact file and line references, ready to paste into a ticket or PR comment. ## Core Features & Use Cases - Five rule categories: Scans for DRY violations (duplicate blocks, copy-pasted validation), KISS issues (deep nesting, long methods, unnecessary abstractions), YAGNI problems (unused methods/fields/imports, over-engineering), naming issues, and structure problems (god classes over 300 lines, long parameter lists). - Framework-aware exclusions: Knows not to flag Spring Data R2DBC repositories, MyBatis mapper interfaces, record-based constructor injection, and generated code, reducing false positives. - Uncertainty labeling: Findings that static text analysis cannot prove (e.g., reflectively-invoked methods) are marked as needing human judgment rather than asserted as violations. - Use Case: Before merging a pull request, run the audit on the changed package to catch a 4-level nested conditional and an unused field, then paste the formatted report directly into the PR comment. ## Quick Start Ask the assistant to run a clean code audit on a specific Java package such as src/main/java/com/example/booking/ and review the categorized findings report.
