Recently I’ve noticed that the web version of claude (claude.ai) is doing a lot better than Cursor. Is it Cursor getting worse or has the web version just gotten better?
For the time being I’ve been asking the web version the questions then asking Cursor to implement the logic and double check it.