The current AI doesn't have intellisense to debug. As a developer, we'd have some memory of what we created with the code/functions. But the debug capabilities are like someone who doesn't know how to debug. E.g. it detects an error, I prompted to identify source, tells me what must be done, fixes the code - it says, technically hasn't fixed the code (i.e. didn't commit or I don't know - but codebase isn't changed). doesn't do impact analysis on other parts of code that could have a similar/downstream issue.

I've to run multiple times and spent half my credits this month debugging in constant loops.