Reproduce “What Does "Now" Mean to Your Computer-Use Agent? Stale Screenshots, Page Loads, and Which Check to Run Before a Click” 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
Vasundra Srinivasan
The authors find that page loading, rather than planning delay, is the main stale-click risk on 31 public pages. Waiting for 500 ms of network quiet left 0.46% stale at a median 1.24 seconds. On a held-out benchmark, a pixel check detected 82% of targets with 4% false alarms in 312 ms; this does not establish stale-click rates on commercial pages.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.