Reproduce “SBVUL: Statement-Based Vulnerability Detection” 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
Martin Pinzger, Katharina Stengg, Christian Macho, Stefan Schöberl, Víctor Mayoral Vilches
SBVul identifies exploitable statements in C/C++ functions through two agents: one proposes candidate lines, while another checks control and data flow evidence. It derives function labels from confirmed candidates. On 200 vulnerable/patched pairs, Opus 5.5 (xhigh) achieved statement-level precision of 0.315 and recall of 0.605, but confirmed candidates in 54% of patched functions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.