返回技能列表
opsecmd
由 wulf715 提供
iOS & macOS Development—暂无评分
你的评分
简介
A swift reminder of both human and agent duties regarding operational security.
描述
该技能的功能
A non-executable OPSEC reminder document for both humans and AI agents. Covers basic due diligence around reviewing and trusting scripts before execution in agentic systems. Written in response to observed abuse of agent trust to cause real harm via malicious skill files.
Addresses the specific attack surface of malicious skill files in agentic systems, a gap that general security checklists do not cover.
何时使用
- Reviewing a third-party skill file before letting an agent execute it
- Spot-checking automated scripts in an agentic pipeline before deployment
- Onboarding team members to baseline OPSEC practices for AI agent setups
- Deciding whether to trust a downloaded script when its behavior is unclear
- Establishing a human-readable policy for agents running with minimal supervision
安装
clawhub install wulf715/opsecmd运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。