I often get a lot of good mileage around planning my changes in the chat view. I then often want to continue my work in the compose view. Either as a single-turn composition or by perhaps using the agent mode. Right now I have to copy/paste a bunch of context which is a bit annoying. I wish I had the ability to condition my compose inference calls based on the context within a given chat.
Hey, thanks for the suggestion.
We’ve got some changes coming in a future update around the chat and composer, so keep an eye out for that as it’ll likely improve on the workflow you describe!