描述
专为 OpenClaw 技能打造的安全优先审查工具。在安装来自 ClawHub、GitHub 或其他来源的任何技能前,请务必使用本工具。
安全与治理 / 审核评估
skill-vetter
描述
专为 OpenClaw 技能打造的安全优先审查工具。在安装来自 ClawHub、GitHub 或其他来源的任何技能前,请务必使用本工具。
安全审计
不符合规则:存在需要人工确认的行为
位置:SKILL.md:77
相关片段:- Prompt injection patterns ("ignore previous instructions", "you are now...")
不符合规则:A user might see 'trust-score: 97' and assume the skill was independently verified, when it is simply the publisher's own claim.
位置:SKILL.md
相关片段:The frontmatter self-assigns a high trust score and claims a prior audit. This is unaudited self-assertion by the publisher, not an independently verified rating.
不符合规则:Minimal. The agent will adopt an auditor persona when this skill is invoked, which is the expected behavior for a vetting tool.
位置:SKILL.md
相关片段:The skill assigns a role ('security auditor') and uses a mild imperative ('must vet'). This is standard role-setting for a checklist tool and does not override user intent or change stopping conditions.