What problem does it solve? Raw coverage percentages tell you what code ran, but not which uncovered code is dangerous to change. This Skill combines line/branch coverage with cyclomatic complexity to compute CRAP (Change Risk Anti-Patterns) scores per method, surfacing the complex, under-tested hotspots that block coverage improvement and make refactoring risky. ## Core Features & Use Cases - CRAP Score Hotspot Ranking: Parses Cobertura XML and applies the formula comp² × (1 − cov)³ + comp per method, returning a ranked table of the riskiest methods. - Automated Coverage Collection: Detects the coverage provider (Coverlet or Microsoft CodeCoverage) per test project, adds a provider package when missing, and runs dotnet test with the correct arguments — including mixed-provider solutions. - Coverage Gap and Plateau Diagnosis: Extracts per-method coverage below configurable thresholds and prioritizes gaps as HIGH/MED/LOW to explain why coverage is stuck. - Use Case: A developer's solution sits at 72% coverage and won't move. Run this Skill to discover that three methods with complexity above 15 and zero coverage account for most of the risk, and get a prioritized list of which tests to write next. ## Quick Start Analyze the coverage of my .NET solution in the current directory and show me the top CRAP score risk hotspots.
