Reproduce “APPATCH: Automated Adaptive Prompting Large Language Models for Real-World Software Vulnerability Patching” 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
Yu Nong, Haoran Yang, Long Cheng, Hongxin Hu, Haipeng Cai
APPATCH patches known C vulnerabilities by tracing data and control dependencies around vulnerable operations. It guides an LLM through root-cause analysis, selects cause-matched repair examples, and screens candidate patches with multiple LLMs. On 97 Zero-Day cases, Claude-3.5 achieved 36.46% Correct F1 versus 28.41% for Manual Exemplars; patches were manually judged, and the system assumes known vulnerability locations and CWE types.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.