Reproduce “Meta²: Precomputing Reusable Knowledge for Online Harness Evolution in Long-Horizon Agents” 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
Lingzhi Yuan, Yiguang Cai, Weize Liu, Siqi Yang, Feijiang Han, Yongkyun Lee, Udari Madhushani Sehwag, Furong Huang
Meta² learns reusable guidance for adapting an agent’s harness, then uses a separate online evolver to keep it unchanged or make run-specific edits based on current evidence. Its frozen Meta-Skill is learned offline from outcome contrasts. On Terminal-Bench 2.1 with GPT-5.6-Terra, it reports 81.00% versus 73.33% for Meta-Harness; evaluation covers two terminal and repository benchmarks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.