Reproduce “Beyond Approved Actions: Runtime Validation of Persistent Outcomes in Agent Workflows” 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
Haoran Zhang, Hengtong Zhang, Zhiyu Liang, Yu Yan, Decheng Zuo, Hongzhi Wang
EffectMatch checks whether an agent’s persistent backend changes match application approval before allowing commit and dependent work. It binds approval to its source, current state, and one execution, then compares staged effects at a registered observation boundary. In controlled replays of 12 STATE-Bench tasks, it rejected all 117 mismatches and preserved 36 clean outcomes; results cover tested tasks and boundaries.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.