vostride/agent-qa — reverse-engineered prompt
Reverse engineered prompt
Build me an open source QA tool where I can write end to end tests in plain English for web and mobile apps, then run them from a dashboard or the command line. I want it to feel friendly for humans but still useful for coding agents, with a polished UI, natural language test files, and a simple setup flow so someone can install it, initialize it, open the dashboard, and run a sample test right away.
The important part is that it should try to recover when the UI changes or an action fails, learn from past runs with memory, and reuse successful action plans so future runs are faster and cheaper. Please support sandboxed setup and teardown hooks in Docker for common runtimes like Node, Python, Bun, and Bash, and let me bring my own model provider, including hosted and local options. Include a small demo project and make the whole thing reviewable as normal code and config. If you need details, look up the current docs online and match that experience.
Want more depth? Deep Reverse