The Passmark project is an open-source AI-powered regression testing framework built on top of Playwright that enables developers to write end-to-end browser tests using natural language instead of traditional scripting. It is designed to simplify and accelerate testing workflows by allowing AI models to interpret human-readable instructions and translate them into executable browser actions. One of its defining features is a cache-first execution model, where AI is used initially to discover how to perform actions, and those actions are then stored and replayed at native speed in future runs. This approach significantly reduces reliance on AI during repeated test executions, improving both performance and cost efficiency. The framework also incorporates multi-model validation, using multiple AI systems to verify assertions and reduce the risk of incorrect results.
Features
- Natural language test execution powered by AI models
- Cache-first execution with Redis-backed step replay
- Auto-healing tests that adapt to UI changes automatically
- Multi-model assertion validation with consensus logic
- Dynamic test data generation and cross-test state management
- Built-in email testing with OTP and content extraction