英伟达发布 SkillSpector：面向 AI 代理技能的开源安全扫描器
英文摘要
NVIDIA has open-sourced SkillSpector, a specialized security scanner designed to analyze AI agent skills for vulnerabilities, malicious patterns, and security risks. The tool aims to help developers and security teams automatically detect threats within agentic AI implementations. Available on GitHub, it addresses the emerging need for safety assurance in autonomous AI systems.
中文摘要
英伟达开源了 SkillSpector，这是一款专门用于检测 AI 代理技能中漏洞、恶意模式和安全风险的安全扫描器。该工具旨在帮助开发者和安全团队自动发现代理人工智能实现中的威胁。它已在 GitHub 上发布，回应了自主 AI 系统日益增长的安全保障需求。
关键要点
NVIDIA has open-sourced SkillSpector, a security scanner targeting AI agent skills.
英伟达开源了针对 AI 代理技能的 SkillSpector 安全扫描器。
The tool detects vulnerabilities, malicious patterns, and general security risks in skill implementations.
该工具检测技能实现中的漏洞、恶意模式及一般安全风险。
It is available on GitHub to support secure development of agentic AI systems.
已在 GitHub 上发布，以支持代理 AI 系统的安全开发。