Reproduce “Why Do Agent Teams Fail Across Boundaries? A Failure Taxonomy for Agent-to-Agent Systems with Private States” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sarath Shekkizhar, Prathyusha Jwalapuram, Zixuan Ke, Vaidehi Patil, Jiayu Wang, Srijan Bansal, Trong Huy Nguyen, Dingfeng Jiang and 7 more
The authors introduce A²-Failures, a taxonomy of agent-to-agent failures derived from more than 2,500 runs across 488 scenarios. It groups 20 failure modes into four categories. The study finds agents may fail to contact needed peers or verify their claims, while task scores can overlook conduct that harms a principal. These findings describe the evaluated corpus.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.