CUADebug: Diagnosing and Repairing Computer-Use Agent Failures
Published in EMNLP 2026, 2026
CUADebug studies failures in computer-use agents and introduces a framework for diagnosing and repairing them through a CUA-specific taxonomy, benchmark, and tool-augmented debugger.
Recommended citation: Weijia Zhang, Kunlun Zhu, Zeyi Liu, Yinting Chen, Tianyi Ma, Jiateng Liu, Jiaxun Zhang, Bingxuan Li, Pan Lu, Xiangru Tang, Heng Ji, and Jiaxuan You. (2026). "CUADebug: Diagnosing and Repairing Computer-Use Agent Failures." EMNLP 2026.
Download Paper
