KoboldCpp v1.84: native RAG with embedded ChromaDB, no separate servers

KoboldCpp is a tool for running AI models locally, widely used for storytelling and role-playing games. With version 1.84, it adds an important feature: you can give it your own documents (PDFs, text files, notes) and it indexes them so the model can search through them during conversation. When you ask a question, it automatically retrieves relevant passages from your documents and includes them in the context. No additional server to install — everything is bundled in a single executable file.