Automation of activities in the browser. Is it possible to implement this and how?

Question

safari

Greetings to all who read these lines.
Is it possible to write an extension for Safari that will move itself in its own window. That is, I need to solder element recognition (OpenCV) there, clicks were simulated, the "Active window" was simulated in each browser tab, of which there will be about 15-18 in each window. And so that in each window of the same Safari there are different actions that I could coordinate separately. In general, is it possible to write this and how? Prompt a direction in which to dig.
Heard about all sorts of Macros-iMacros, but this will not work. It’s not enough just to program clicks in certain places, you also need to make the program understand where to click and what to follow.
Something like java.awt.Robot in JAVA, but more fancy.
PS sorry for my sophisticated whims.

Answer 1

https://github.com/SeleniumHQ/selenium/wiki/Safari...