You guys are getting way better with user sentiment matching product and this is very clear in the fact that we now just have an agent that does everything and is a chat. However the idea of Composer was that it naturally read through your entire code base and was your AI assistant for your whole project. While agent is very useful it is not an AI assistant for a whole project. It is only for one chat and really only good for about 10 messages in that chat which is proven by the fact that it is asking us to Close up that chat and use it as context for a new chat because having a ton of Messages in a chat makes the response as less effective.
However, the obvious solution to this is to make one central composer that you can chat with about the scope of your entire project and then have chats that function like the current agent chat work for Each individual feature. I’m good enough with AI software development at this point that I could do this in a day. So if you guys can’t do it in a week, I am asking and begging please to make the software open source because it would just increase my productivity so much.
To be very clear, what I am asking for is a main chat that you can chat with and talk about the context of your project and everything you want it to do and all the documentation necessary and then have multiple agent chats that function like the current agent chats that can not only do everything the current agent chat does but are smart about summarizing themselves. As context for new chats.
This way you would not have to make composer/Agent chats over and over and over again to get one issue done.