Llm agents benchmarked on assessing software supply chain vulnerabilities
VEX-Bench: Benchmarking LLM Agents for Assessing Exploitability of Software Supply Chain Vulnerabilities
Cryptography and SecurityMachine LearningSoftware Engineering
Summary
Software projects depend on other software packages, but some of those packages can have security problems that might be exploited by hackers. Figuring out if these vulnerabilities can actually be used to attack a project is hard and usually done manually. The authors created VEX-Bench, a test set that measures how well large language model agents can decide if a known vulnerability in a dependency can affect a project. They tested several models and found that while some do well at saying if a vulnerability is exploitable, fewer can explain in detail why. This helps highlight the challenges in automating this important security task.
What this means in practice
- •For software security teams: Evaluate and improve tools that analyze whether known vulnerabilities in dependencies truly threaten their projects.
- •For software package maintainers: Prioritize fixing or patching vulnerabilities in dependencies that actually impact downstream projects using advanced automated assessments.
Authors
Jiahao Shi, Edward Tsien, Yifeng Di, Hongjiao Zhang, Yuan Tang, Ronit Dey, Ilona Shishov, Gal Netanel, Zvi Grinberg, Vladimir Belousov, Bat-Zion Rotman, Ilan Pinto, Tianyi Zhang
Abstract
The software supply chain has become an increasingly exposed attack surface because of its reliance on intricate yet fragile dependencies. Existing defenses such as GitHub Dependabot often raise many false alerts because their coarse-grained matching cannot determine whether a vulnerable dependency is actually exploitable. Security analysts typically spend substantial time assessing vulnerability exploitability case by case. Recent LLM agents have emerged as promising candidates for this task given their advanced capabilities in coding and cybersecurity, yet no existing benchmark evaluates them on it. Prior benchmarks target zero-day settings, where agents detect and exploit previously unknown vulnerabilities. In contrast, software supply chain security focuses on how known vulnerabilities in upstream dependencies affect downstream projects. This requires agents to reason across repositories and determine whether an upstream vulnerability is exploitable in the downstream project. To address this gap, we introduce VEX-Bench, the first benchmark for evaluating LLM agents' ability to assess the exploitability of software supply chain vulnerabilities. It contains 75 real-world cases mined from GitHub and labeled by security experts, covering Python, Java, and Go. We evaluate nine models across three agent harnesses. While GPT-5.5 and Claude Opus 4.6 reach approximately 80% F1 on binary vulnerability-status classification, only GPT-5.5 surpasses 70% macro-F1 on fine-grained justification classification. This gap highlights the challenge of moving beyond binary exploitability assessment to identifying fine-grained exploitability reasons. Code and data: https://github.com/steven1518/vex-bench