Reproduce “EvidenT: An Evidence-Preserving Framework for Iterative System-Level Package Repair” 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
Chenyu Zhao, Minghua Ma, Shenglin Zhang, Zeshun Huang, Yongqian Sun, Chetan Bansal, Saravan Rajmohan, Dan Pei
EvidenT repairs failed system-package builds by carrying build logs, prior repair outcomes, package analyses, and architecture knowledge across iterations, then validating artifact-level edits through OBS. In a study of 219 failed RISC-V packages, GPT-5-mini with EvidenT reached OBS’s succeeded state for 118 packages (53.88%), compared with 42 (19.18%) without preserved evidence; build success does not establish runtime correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.