arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.37901cs.CRcs.AI
收藏

安装前扫描:检测与缓解技能中的局部及复合漏洞

Scan Before You Install: Detecting and Mitigating Local and Composite Vulnerabilities in Skills

organization= Stein Faculty of Computer;Information Science, Ben-Gurion University of the Negev , city= Beer-Sheva , country= Israel

Parth A. Gandhi, Avital Aviv, Prasanna N. Wudali, Yuval Elovici, Asaf Shabtai

AI总结:

本文针对现有技能扫描器的不足,开发六模块扫描器,可检测技能局部与复合漏洞,在多数据集上性能优于同类工具,且提出的缓解措施能有效阻止攻击并保留合法功能。

摘要:

智能体技能是基于大语言模型(LLM)的智能体为完成任务加载的自然语言指令、脚本和参考文件的集合,可在用户授权权限下访问文件系统、shell和网络,其存在的漏洞会导致不安全执行。现有扫描器主要针对恶意技能,未关注公共仓库中技能的意外漏洞,且孤立分析技能,忽略多技能间的交互。为解决这些问题,本文开发了包含内容提取、局部分析、复合分析、验证、报告和缓解的六模块扫描器。局部分析用于检测良性技能中的意外漏洞:在SkillTrustBench的556例排行榜子集上,其召回率达95.90%,误报率为4.22%;在1000个MalSkillBench技能的平衡样本上,其F1值为0.947,优于Cisco Skill Scanner、NVIDIA SkillSpector、AgentSkillsScanner和SkillSieve。在来自75个星标最多的GitHub仓库的1587个技能中,三名分析师在23个仓库中确认了59个发现中的53个(准确率89.8%)。本文为每个漏洞开发了概念验证(PoC)攻击,向维护者披露了10个中高严重程度的案例,并对其余案例测试了缓解措施。复合分析通过共享资源、调用和指令介导的交互建模技能间的依赖关系:在SCR-Bench上,其F1值为0.991(准确率98.7%,召回率99.6%),而CompoSkill发布的三种SkillsGuard配置的F1值为0.816-0.858;其在GitHub上识别出两个复合发现,经PoC攻击验证。缓解模块推荐技能替换、补丁和配置更改:在55个GitHub发现中的47个上,缓解措施在所有试验中阻止了PoC攻击,同时保留了测试的合法功能。

英文摘要:

An agent skill packages natural-language instructions, scripts, and reference files that an LLM-based agent loads for a task. Skills may access the file system, shell, and network under user-authorized permissions; weaknesses can lead to unsafe execution. Existing scanners primarily target malicious skills, not unintended vulnerabilities in public repositories, and analyze skills in isolation, overlooking multi-skill interactions. To address these gaps, we developed a six-module scanner: Content Extraction, Local Analysis, Composite Analysis, Validation, Reporting, and Mitigation. Local Analysis detects unintended vulnerabilities in benign skills. On a 556-case leaderboard subset of SkillTrustBench, it achieved 95.90% recall with a 4.22% false positive rate; on a balanced sample of 1,000 MalSkillBench skills, its F1 was 0.947, outperforming Cisco's Skill Scanner, NVIDIA's SkillSpector, AgentSkillsScanner, and SkillSieve. In 1,587 skills from the 75 most-starred GitHub repositories, three analysts confirmed 53 of 59 findings (89.8% precision) across 23 repositories. We developed proof-of-concept (PoC) attacks for each vulnerability, disclosed 10 medium- and high-severity cases to maintainers, and tested mitigations on the remaining cases. Composite Analysis models dependencies among skills through shared resources, invocations, and instruction-mediated interactions. On SCR-Bench, its F1 score was 0.991 (98.7% precision, 99.6% recall), versus 0.816-0.858 for three SkillsGuard configurations released with CompoSkill. It identified two composite findings on GitHub, validated by PoC attacks. The Mitigation module recommends skill substitution, patches, and configuration changes. For 47 of 55 GitHub findings, mitigations prevented PoC attacks in all trials while preserving tested legitimate functionality.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑