From my experience the tests seem to break extremely easily and there is more time spent fixing them than finding actual bugs in the UI.
Great for sanity checks (i.e. did everything deploy correctly, are my DNS settings correct, is there no render-blocking bug somewhere, etc.), but beyond that it’s too brittle to write an extensive test suite in.
Its good for some things, but there’s an almost unlimited number of things you could test with the UI, and UIs change frequently. Its def more of a burden than a help in most cases.
My experience as well is that it quickly become a burden, ref the Ice Cone antipattern.
A general programming discussion community.
Other programming communities: