crap-analyzer

To find risky code changes in a pull request, you compute Change Risk Anti-Patterns (CRAP) by analyzing a unified diff. This ranks recently changed functions worst-first by combining diff-scoped complexity and test coverage signals to surface under-tested paths.

The CRAP formula used for risk scoring is CRAP(m) = comp(m)^2 × (1-cov(m))^3 + comp(m). It calculates risk by combining a function's complexity with its test coverage deficiency to identify refactoring targets.

Yes, you can run diff analysis and risk scoring on Python or Java/Kotlin repositories. The analyzer supports TypeScript/JavaScript, Python, Java/Kotlin, Go, Ruby, C#, Rust, and PHP across PR and branch workflows.

To generate test stubs for complex functions flagged by mutation testing, the CRAP analyzer turns each finding into a targeted test stub and a concrete refactor proposal. It applies rules to avoid unsafe auto-changes while planning fixes.

If test coverage files are missing during refactor planning, the analysis runs with coverage set to 0% and clearly flags the limitation. It auto-discovers compatible coverage formats when available to ensure accurate risk scoring.