Hey HN,
This weekend our team created Playwright-AI. It's a tool that hooks Anthropic's computer use model into Playwright (on top of Puppeteer) to execute browser tests from natural language test cases.
ai(”Test users can create a new project.”)
Why tho? Testing Sucks! We wanted to see if we could just feed test cases in natural language to AI and it would test them exactly as a human would do. Writing E2E UI tests is super brittle, you change `sign in` button to say `login` and everything breaks…! If you've written UX E2E tests in a large org I'm sure you will know this pain well. We were inspired by zerostep.com
Fully Open-Source: MIT-licensed, free to use, and open for contributions.
This is just the initial version, and we're eager to learn and make it even better, based on community feedback.
