Hey Jacob. Thanks for the reply. Probably a better description of what I meant is more like what you see in ide conflict merge workflows. The problem with git branch code reviews is how heavy they are. With ai writing code, the velocity is so much higher that jumping out to another tool is just too cumbersome. Also the code reviews feedback doesn’t flow back into the v0 agent. What I’m really hoping for is something where I can accept some changes, reject others, and give prompted direction when I reject, and have v0 address those comments. So think of it as a human in the loop process to any of the code modification steps in the v0 answer chain. The user can toggle this on, and as changes are made, the user can either accept, reject without comments, or reject with comments and in any case of a reject, v0 continues work with that rejection and added comment as additional context before fully completing the answer chain. With this I think you would address a lot of the users concerns around v0 editing code it shouldn’t, or just going down a rabbit hole that didn’t reflect the users original intention. Plus LLMs react pretty well to additional context.
