yes I have been facing this issue and all aspects of task execution has become lower in quality, I have moved to opencode.ai for all my routine repeated tasks for now -- a bit slow but works well on the free model, looking for alternative for difficult questions to run in terminal