With the advancement of artificial intelligence technology, the field of E2E (end-to-end) testing is experiencing a new wave of innovation. Tools like Midscene.js, recently open-sourced by the ByteDance web-infra team, and Shortest, developed by independent developers, are changing traditional testing methods.
Midscene.js utilizes a multimodal large language model that can intuitively "understand" user interfaces and perform corresponding actions. The main feature of this framework is its convenient browser plugin, which allows users to interact with web pages using natural language without writing any code. This plugin supports three core functionalities: interaction, extraction, and assertion, significantly simplifying the testing process.
At the same time, Shortest, developed by a startup team, also demonstrates the potential of AI in the testing field. By combining natural language processing and interface screenshot analysis, Shortest can directly generate E2E test cases, significantly reducing the time teams spend on repetitive testing tasks.
Industry experts point out that as AI's coding and multimodal capabilities mature, the automation level of basic E2E testing scenarios has reached a considerable degree of sophistication. The emergence of these innovative tools not only improves testing efficiency but also promotes the optimization of the entire software development process.
Open source address: https://github.com/web-infra-dev/midscene
