1 paper · 1 filter
Zimu Wang, Xiaobo Wang, Siyu Ye +11
Agent-evolution methods use execution feedback to improve prompts, skills, tools, and orchestration, but their implementations often rebuild the same experimental machinery. We pre…