Reproduce “World Editing: Intervening on Executable Worlds at Increasing Depth” 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
Max Ku, Nok-Kan Law, Yu-Chien Tang, Shih-Ying Yeh, Ping Nie, Andy Zheng, Tat Hei Lai, Fei-Yueh Chen and 10 more
The authors introduce IGMWorld, an environment for editing and running game mods, and IGMBench, which evaluates world editing in Minecraft and Terraria across 110 tasks. Agents modify existing games and are assessed on state and behavior. GPT-5.6 Sol with Codex-CLI passes all criteria on 78.2% of tasks, but success falls from 96.3% at L1 to 48.1% at L4; these are functional results, not visual validation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.