characterization-test-generator

To generate characterization tests for existing Swift code, you use a tool that discovers your code via glob patterns and public API surface, then produces test templates that freeze the current runtime behavior as a safety net. This captures pure computations, state mutations, and async operations.

A characterization test documents the actual runtime behavior of existing code rather than its intended design. You need it when refactoring legacy codebases or planning AI-assisted changes, providing a regression safety net by freezing current behaviors before modifications.

Yes, characterization tests can freeze async operations and side effects in Swift. The generator provides specific templates for these behaviors, alongside patterns for pure computations, state mutations, error paths, and edge cases, ensuring comprehensive behavior capture.

The best way to create a regression safety net for AI-assisted changes is to generate characterization tests that document current behavior. By discovering code through file globbing and API surface detection, you freeze existing logic across multiple behavior templates before applying modifications.

Yes, the characterization test generator works without external dependencies. It operates independently to discover your Swift codebase, classify behaviors, and generate test templates for pure computation, state mutation, async operations, side effects, errors, and edge cases.

The templates generated for edge cases and error paths document specific runtime behaviors to prevent regressions. The generator classifies and provides templates covering pure computation, state mutation, async operations, side effects, error paths, and edge cases for comprehensive legacy code coverage.