Import Analysis Fails at Runtime: Why Coverage-Based Test Selection Wins

Vitest and Jest's built-in test selectors use static analysis to avoid running every test on each change, but they fail to detect tests that exercise dynamically loaded modules—allowing real bugs to slip through CI. When code loads modules via runtime-computed paths (registries, dependency injection, feature flags), static tools see no connection between test and source. Coverage-based test selection fixes this by tracking which tests actually executed which code, catching issues that static analysis would miss.