AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution Traces Paper • 2608.23041 • Published 12 days ago • 64
K-BrowseComp: A Web Browsing Agent Benchmark Grounded in Korean Contexts Paper • 2606.02404 • Published Jun 1 • 60
Why Far Looks Up: Probing Spatial Representation in Vision-Language Models Paper • 2605.30161 • Published May 28 • 60
Personalize-then-Store: Benchmarking and Learning Personalized Memory for Long-horizon Agents Paper • 2605.25535 • Published May 25 • 46
Safe and Scalable Web Agent Learning via Recreated Websites Paper • 2603.10505 • Published Mar 11 • 28