In-between adjustments/clarifications of plans and agent responses

Feature request for product/service

Cursor IDE

Describe the request

Often there are small details in an answer that are not 100 percent accurate. It is always tedious to read everything first and then adjust the nitty gritty details at the end. It would be nicer if you could immediately highlight text passages and then add annotations to them, for example, so that you can immediately correct “wrong” or not entirely “correct” answers. In my opinion, this would make the whole process much easier to control and I suspect it would lead to better results, as the context is also more clearly specified for the LLMs with the highlighting.