Reproduce “LegoEnv: Self-Evolving Agentic Environments via Modular Recombination” 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
Zhuosong Jiang, Chengxu Wu, Junqi Liu, Yujie Yi, Siqin Wang, Zhixuan Huang, Jinqian Jiang, Lijun Xiao and 6 more
LegoEnv builds business workflow environments from reusable intents, rules, datasets, and application specifications. It checks verifiers against initial and reference states, then diagnoses and repairs verifier or world faults before resampling. In round two, yield—in this case, environments with an accepted trajectory in eight teacher attempts—increased from 71.6% to 88.3%. Qwen3.6-27B SFT scored 16.0% to 32.0% on AutomationBench; sound unsolved tasks remained unchanged.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.