I like debug mode - it proposes hypothesis, inserts loggers, analyse logs, and fix the issue. However, it happens to me very often that the fix being applied is ad-hoc and does not work. I must say the bugs I’m facing to are quite difficult. From the chat I can see that the fix is propose prematurely, the root cause is not known yet and further investigation is needed, but it decides to apply the fix. This is annoying, because then I have to ask to revert and add further implementation and it is happening repeatedly….
It would be awesome to somehow combine debug and plan mode… it would be debugging and creating the plan, until the plan is reliable, then the user could decide to fix it.
I see that the proposed approach is too heavy for simple bugs (e.g. typos), however, when facing a difficult bug, the debug mode does not work very well.