1 paper
Jinghan Xu, Yikai Zhang, Aili Chen +3
Agent harnesses shape how language-model agents use instructions, tools, and runtime components, but adapting these harnesses requires costly verification. Existing propose-and-ver…