Scaffolding Dexterous Manipulation with Vision-Language Models | Read Paper on Bytez