characterization-test-generator

To generate characterization tests for legacy code, you document the current behavior of untested modules using templates for pure computations, state mutations, and async operations to lock in a regression suite before refactoring.

Creating a safety net for AI-assisted refactoring involves generating a runnable characterization test suite that records current code outputs and side effects, allowing you to detect unintended changes immediately after modifications.

Documenting current behavior in untested modules across languages requires generating explicit characterization templates that capture actual outputs, state mutations, and error paths, producing a runnable test suite that validates existing functionality.

Yes, characterization testing works for modules with async operations and side effects by utilizing specific test templates designed to capture and lock in the current behavior of these complex execution paths against regressions.

You should use characterization tests instead of traditional unit tests when dealing with legacy or untested code where the actual current behavior needs to be documented and preserved as a regression safety net prior to refactoring.

Locking in current behavior before changing legacy code is necessary because characterization tests capture the actual outputs and side effects, ensuring that your refactoring efforts do not inadvertently alter existing functionality.