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.
Can Agents Trust Their Skills? Uncovering Unsafe Chains of Trust in Skill-Based LLM Agents
Institute of Information Engineering, Chinese Academy of Sciences(中国科学院信息工程研究所);Worcester Polytechnic Institute(伍斯特理工学院);School of Cyberspace Security, University of Chinese Academy of Sciences(中国科学院大学网络空间安全学院)
When Prompts Become Payloads: A Framework for Mitigating SQL Injection Attacks in Large Language Model-Driven Applications
Hasso Plattner Institute, University of Potsdam, Germany(波茨坦大学霍斯曼-普拉特研究所, 德国);Chemnitz University of Technology, Chemnitz, Germany(Chemnitz技术大学, Chemnitz, 德国)
ETDI: Mitigating Tool Squatting and Rug Pull Attacks in Model Context Protocol (MCP) by using OAuth-Enhanced Tool Definitions and Policy-Based Access Control
Amazon Independent Researcher(亚马逊独立研究者);Project Kuiper Security (KPES)(项目Kuiper安全);Proactive Security OWASP, Amazon Web Services(主动安全OWASP,亚马逊网络服务);Intuit Adversarial AI Security reSearch (A2RS)(Intuit对抗AI安全reSearch)
本文提出 ETDI 安全扩展,通过密码学身份验证、不可变版本化工具定义、OAuth 2.0 权限管理及基于策略的细粒度访问控制,缓解 MCP 中的工具投毒和 Rug Pull 攻击,构建更安全可控的 AI 工具交互生态。
Shadow in the Cache: Unveiling and Mitigating Privacy Risks of KV-cache in LLM Inference
State Key Laboratory of Blockchain and Data Security(区块链与数据安全国家重点实验室);Zhejiang University(浙江大学);Huawei Technology(华为技术);Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security(杭州高新技术区(滨江)区块链与数据安全研究院)
SafeMind: Benchmarking and Mitigating Safety Risks in Embodied LLM Agents
Brain-inspired Cognitive AI Lab, Institute of Automation, Chinese Academy of Sciences(脑启发认知AI实验室,自动化研究所,中国科学院);Beijing Key Laboratory of Safe AI and Superalignment(北京安全AI与超对齐关键实验室);Beijing Institute of AI Safety and Governance(北京AI安全与治理研究所);School of Artificial Intelligence, University of Chinese Academy of Sciences(人工智能学院,中国科学院大学);Long-term AI(长期AI)
Mitigating Deep Learning Vulnerabilities from Adversarial Examples Attack in the Cybersecurity Domain
Institute of Electrical and Electronic Engineers(电气与电子工程师协会)
已知深度学习模型通过在具有最佳准确度的大型数据集上采用多个历元和训练样本来解决分类和回归问题。然而,这并不意味着它们是防攻击的或不暴露于漏洞。新部署的系统,特别是在公共环境(即公共网络)上的系统,容易受到来自各种实体的攻击。此外,已发表的关于深度学习系统的研究(Goodfellow et al.,2014)已经确定了大量的攻击点和广泛的攻击面,这些攻击面有证据表明存在来自对抗性示例的利用。成功利用这些系统可能会导致严重的现实世界的影响。例如,(1)对运行深度强化学习系统的自动驾驶汽车的对抗攻击会对人类产生直接的错误分类,从而导致不幸的事故。(2)自动驾驶车辆误读红灯信号可能导致汽车撞上另一辆汽车(3)将行人车道误分类为交叉路口车道,可能导致车祸。这只是冰山一角,计算机视觉的部署并不完全集中在自动驾驶汽车上,还包括许多其他领域,这将对现实世界产生决定性的影响。必须在发展的早期阶段减轻这些脆弱性。在实际部署之前,必须在全球范围内制定和实施基线安全标准。
Understanding and Mitigating Token-Pruning-Induced Vulnerabilities in VLMs
University of Electronic Science and Technology of China(电子科技大学);Southwestern University of Finance and Economics(西南财经大学);Tongji University(同济大学);Shanghai Innovation Institute(上海创新研究院)
Unveiling and Mitigating Backdoor Vulnerabilities based on Unlearning Weight Changes and Backdoor Activeness
The Hong Kong University of Science and Technology (Guangzhou)(香港科技大学(广州));The Chinese University of Hong Kong, Shenzhen(香港中文大学(深圳));Shenzhen Research Institute of Big Data(深圳大数据研究院)
Virtualization-based Penetration Testing Study for Detecting Accessibility Abuse Vulnerabilities in Banking Apps in East and Southeast Asia
Singapore Management University(新加坡管理大学);MIE-SPPU Institute of Higher Education(MIE-SPPU高等教育学院);i-Sprint Innovations Pte. Ltd(i-Sprint创新私人有限公司);Promon(普罗蒙)