AgentBeats: Agentifying Agent Assessment for Openness, Standardization, and Reproducibility

Xiaoyuan Liu, Jianhong Tu, Yuqi Chen +26

Agent systems are advancing quickly across domains, but their evaluation remains fragmented. Most benchmarks rely on fixed, LLM-centric harnesses that require heavy integration, cr…