1 paper
Yimo Lin, Zhen Zhang, Yibin Li
While expert-validated "LLM + script" workflows deliver significant value, they remain static: they encode hard-won domain knowledge yet fail to adapt execution based on feedback.…