Large language model (LLM) agents increasingly operate over long-horizon interactions involving tool use, persistent state, evolving authorization, and exter…
机构:IBM
来源:arXiv 2609.16305 | AI4Papers 论文推荐平台