📄 中文摘要
在 ClawHub 上发现了 824 个恶意技能，占总数的 20%。使用 OpenClaw 并从社区安装技能的用户需要学会识别不良技能。作为一名拥有 20 年经验的网络安全专业人士，作者分享了自己在审计 AI 代理技能时的简化流程，提供了一种 5 分钟内可执行的检查方法，帮助用户快速识别潜在的安全风险，确保其使用的技能不会对个人信息造成威胁。
如何在 5 分钟内对 OpenClaw 技能进行安全审计（防止其窃取你的密钥）
📄 English Summary
How to Security-Audit an OpenClaw Skill in 5 Minutes (Before It Steals Your Keys)
The article highlights the presence of 824 malicious skills on ClawHub, accounting for 20% of its total offerings. Users of OpenClaw who install community skills must learn to identify harmful ones. A cybersecurity professional with over 20 years of experience shares a simplified process for auditing AI agent skills, providing a quick 5-minute check that users can perform to identify potential security risks and ensure that the skills they use do not threaten their personal information.