by 安全扫描
OpenClaw
安全
high confidence该技能是一个仅用于指令的 transformer，用于重新格式化并分类用户粘贴的邮件内容；其请求的资源与行为均与此目的相符。
评估建议
This skill is instruction-only and internally consistent with its stated purpose. Before installing or using it: (1) test with non-sensitive sample emails to confirm the strict output format meets your needs; (2) avoid pasting confidential or personal data if your agent uses a remote model or third-party API (the SKILL.md does not control which model or logging policy is used); (3) remember the "Prompt:" line is a best-guess and may be incorrect — verify before acting on assumptions; (4) if you ...详细分析 ▾
✓ 用途与能力
Name/description (reverse AI in email, extract intent) align with the SKILL.md instructions; no unrelated binaries, credentials, or config paths are required.
ℹ 指令范围
SKILL.md is self-contained and only describes how to parse and reformat provided email text. It does not instruct reading system files or external endpoints. Note: the skill processes whatever email text the user pastes, which may include sensitive PII — this is expected for an email-parsing tool but is a privacy consideration when using remote models or shared agents.
✓ 安装机制
No install spec or code is included (instruction-only skill). Nothing is written to disk or downloaded by the skill itself.
✓ 凭证需求
No environment variables, credentials, or config paths are requested. The README mentions the agent will use the user's configured model, which is appropriate and proportional.
✓ 持久化与权限
Flags are default (always:false, user-invocable:true, model invocation allowed). The skill does not request persistent/system-wide privileges or modify other skills.
安全有层次，运行前请审查代码。
运行时依赖
无特殊依赖
版本
latestv0.0.52026/4/20
- 新增明确规则：项目符号须陈述实际内容，不得描述发送者行为。 - 新增规则：禁止在项目符号列表中使用“Bullet:”或“Point:”等前缀元标签。 - 无其他实质性更改；本版本进一步阐明输出格式要求。
● 无害
安装命令
点击复制官方npx clawhub@latest install defluff
镜像加速npx clawhub@latest install defluff --registry https://cn.longxiaskill.com