Today, I experimented with an open-source project called CUA (Computer-Using Agent) Project Browser-use, which allows AI to directly control a local browser. It has been gaining stars quickly on GitHub.
https://github.com/browser-use/browser-use
🌐 What is Browser-use?
enabling automated web operations, such as:
(clicking, typing, scrolling, etc.) (suitable for data analysis, automated reporting, etc.) (can achieve functions similar to automatic form filling and searching with ChatGPT).
that can be tried directly, saving the hassle of local configuration.
🛠 Try running code: Experience Browser-use
making it more secure and flexible.
For example, I gave it the task: Help me book a flight ✈️
In the end, all I had to do was input my username and password and make the payment.
Below are some of the AI's operational processes.
🤖 Official demo
Job hunting👷🏻♂️
Prompt: Read my CV & find ML jobs, save them to a file, and then start applying for them in new tabs, if you need help, ask me.
Downloading model files📃
Prompt: Look up models with a license of cc-by-sa-4.0 and sort by most likes on Hugging face, save top 5 to file.
💡 Give a shoutout to Windsurf!
Job hunting👷🏻♂️
Prompt: Read my CV & find ML jobs, save them to a file, and then start applying for them in new tabs, if you need help, ask me.
Downloading model files📃
Prompt: Look up models with a license of cc-by-sa-4.0 and sort by most likes on Hugging face, save top 5 to file.
. There are a few things that are really great:
: No need to deal with Python versions, NPM installations, or other messy stuff; everything is set up automatically. 
: When I wanted to adjust the Browser-use code, I just let Windsurf modify it automatically instead of spending half a day flipping through documentation and debugging. 
: For example, default user information and API keys can be filled in automatically, saving a lot of trouble. 
With Windsurf, the frustration of coding is reduced by a large margin, truly a "blessing for those obsessed with coding" 😂.