Reproduce “GitHarness: Git Init Your Harness Working Memory for Perpetual User Requirements” 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
Zhibang Yang, Xinke Jiang, Yuxuan Liu, Mingyu Zhang, Zhixin Zhang, Zhengxing Song, Yue Fang, Guohong Qiu and 4 more
GitHarness pairs each requirement with a recoverable work checkpoint, then selects a compatible historical version before resuming a task-native executor. On StackPlanner and DeepSeek-V4-Flash SWE-bench Verified code tasks, removing historical version selection reduced resolved rate from 74.0 to 44.0 and increased average token use from 1,291.4K to 4,513.5K. These results are from benchmark tasks, not real-world deployments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.