What problem does it solve? Hand-written code quality reports contain roughly 10% errors in their first drafts: mismatched line-count totals, missed aggregated test files, header-versus-appendix file count contradictions, and double-counted subdirectories. This Skill adds four verification gates that catch these errors before a report reaches the user. ## Core Features & Use Cases - Statistical Formula Consistency: Cross-checks the sum of per-directory line counts against a single unified find ... | wc -l total, flagging any mismatch instead of trusting manual addition. - Aggregated Test File Detection: Uses grep -rhoP 'class\s+\w+Tests\b' to find test classes inside aggregated files (e.g., DomainModelTests.cs) that simple filename matching (FooTests.cs ↔ Foo.cs) misses. - Scope and Boundary Verification: Reconciles the report header's claimed file count with the appendix table, and separates top-level directory counts from subdirectory counts using find -maxdepth 1. - Use Case: After drafting a code quality report with an Appendix A file statistics table and test coverage data, run these four gates to confirm every number is reproducible from shell commands before delivering the report. ## Quick Start Ask the agent to run the four quality gates on the completed code quality report and fix any statistical or coverage discrepancies before delivery.
