FEATURE REQUEST Context Management And Multi Project Sessions

[Mostafa_s_Academy1]

I want to clarify that this feature is essential and must be given attention due to the advancements of AI models. The feature I request is: 1. Context Compression: Gemini gets a summary of the conversation after reaching a certain amount of tokens. The second feature is session persistence. In the session, you can start new chats on the same projects. The integrations remain the same, everything remains the same, but the session is different. This gives Gemini fresh context and will allow it to work better and better. 3. Persistent memory: This one is optional by the user. Gemini can remember other projects and search the user’s chats on demand. How this will benefit: Say I have an automation or a similar idea I want to redo; I can just tell Gemini that and search the chat because it has a persistent memory, so it will be better. 4. Codebase Indexing: The AI automatically indexes the codebase, big or small, and is able to manage it effectively. There are some other features, like hardening the way Gemini is instructed, because for some reason I find the AI misleading – not because it’s a liar. I tested other instances of Gemini outside of Google AI, and they work pretty well, so I feel that the instructions need an update. Because, from what I feel, the instructions told the AI to be very, very, very helpful, so much so that the AI became a ‘yes man’ most of the time. (‘Yes! Everything is fixed! Everything is alright! The code is 100 percent production-ready!’ while it’s really not.) I hope this one really gets fixed by the release of Gemini 4. The last thing is implementing a skills system: downloading GitHub skills so that the AI can do tasks the same way always.