browser-use/web-ui ? reverse-engineered prompt

Build me a simple web app for running an AI agent in the browser.

I want a clean WebUI where I can type a task, pick an AI model, and watch the agent browse websites and complete the job for me. It should support a few popular providers like OpenAI, Anthropic, Google, Azure OpenAI, DeepSeek, and local Ollama models, and I should be able to set my API keys in an env file.

Please make it so I can use my own Chrome browser session too, so I do not have to log in again on websites, and let me keep the browser open between tasks if I want to see what happened. If possible, include a way to run it locally with Python, and also make it easy to run with Docker.

I’d like the interface to feel friendly and straightforward, with a visible browser view or recording so I can follow along while the agent works. If you need current setup details, check the docs online and make it work cleanly.