Download
Fastbot_Android (Fastbot 2.0) is a model-based automated testing tool by ByteDance designed to discover stability or usability issues in Android apps by modeling GUI transitions rather than relying purely on random interactions. It blends machine learning and reinforcement-learning approaches to build a transition graph of UI states and use that model to intelligently explore possible user interactions — aiming to replicate more human-like usage patterns and uncover hidden bugs, crashes, or edge cases. Compared to traditional random-input tools (like Monkey), Fastbot supports much faster action insertion (up to ~12 actions per second) and can handle a variety of Android OS versions (from older through modern, including customized OS variants). It also supports model reuse: once a model of an app’s GUI transitions is built, subsequent testing sessions can reuse the model to speed up testing or target known risky paths.
Features
- Model-based GUI state transition graph building for intelligent app testing
- Reinforcement-learning–based exploration to prioritize high-reward (likely bug-inducing) paths
- High-speed action insertion (≈ 12 UI actions per second) for scalable fuzz testing
- Compatibility across wide range of Android versions (including modified vendor OS variants)
- Model reuse: ability to save and reload GUI transition models for repeated or regression testing
- Open-source, extensible architecture (Java + C++) that allows custom tuning or enhancements
Project Samples
