Store incoming prompt parts for AI engineers before builds

Your incoming webhook parts arrive unassembled, causing fragmented prompt builds and slowing voice model work. Consolidating them into keyed storage entries gives engineers ready prompts for training and deployment before the next build window.