Newest papers

How Effectively Do Security Skills Guide Coding Agents in Vulnerability Detection in Linux Kernel?

Haoran Xue, Nikta Akbarpour, Gias Uddin, Song Wang · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.25705 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Os agentes de codificação de vulnerabilidades de grande potencial (LLM), como Claude Code, são cada vez mais usados para auditorias de segurança, com habilidades específicas de tarefa que fornecem procedimentos de análise de vulnerabilidades reutilizáveis. No entanto, a sua eficácia na detecção de vulnerabilidades reais e as razões para os seus fracassos permanecem inexplicáveis. Os resultados mostram que 24,4% dos executivos localizam a função correta, 18,3% identificam a função correta e a declaração de segurança responsável usando Claude Code com quatro LLMs subjacentes em 100 vulnerabilidades do kernel real do Linux em dez classes, com um total de 14800 execuções. A nossa análise de fracassos revela limitações na aquisição de evidências, raciocínio de estado-relação, interpretação de código evolutivo, verificação de

Machine translation (Meta M2M100); the English original follows.

Abstract

Large language model (LLM) coding agents, such as Claude Code, are increasingly used for security auditing, with task-specific skills providing reusable vulnerability-analysis procedures. However, their effectiveness in detecting real vulnerabilities and the reasons for their failures remain unclear. We conduct the first large-scale empirical study of 37 security skills using Claude Code with four underlying LLMs on 100 real Linux kernel vulnerabilities across ten classes, totaling 14800 runs. We evaluate vulnerability detection at the function, statement, and vulnerability-class levels. Results show that 24.4 percent of runs locate the correct function, 18.3 percent identify both the function and responsible statement, and 15.7 percent correctly classify the vulnerability. Our failure analysis reveals limitations in evidence acquisition, state-relation reasoning, code interpretation, candidate verification, investigation coverage, and reporting. Based on these findings, we propose SKILL-SIF, a self-improving framework that iteratively evolves security skills by learning from missed vulnerabilities and failure patterns. We evolve five top-performing skills using the original benchmark and evaluate them on 100 previously unseen vulnerabilities. SKILL-SIF improves fully correct vulnerability detection by up to eight times (from 2 percent to 16 percent) on unseen vulnerabilities, while aggregate accuracy across four LLMs increases by 1.85 times (from 6.05 percent to 11.20 percent). These findings highlight the limitations of existing security skills and demonstrate the potential of failure-driven skill evolution to improve vulnerability detection in coding agents.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Vulnerability (computing), Coding (social sciences), Benchmark (surveying), Computer security, Secure coding

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange