Skill Trust Decision
grok-image-tool-pro
纯文档型Skill，声明了exec/write权限但无实际脚本；文档包含大量功能示例代码，但存在文档欺骗风险（权限声明宽泛、功能描述过于宽泛如"自动化执行""命令执行"等）
Why this conclusion was reached
0/4 dimensions flagged Pass
Declared vs actual capability
Declared resources and inferred behavior are broadly aligned.
Review
Hidden execution and egress
1 lower-risk artifacts were extracted and still need context.
Pass
Attack chain and severe findings
There is no explicit malicious chain in the report.
Review
Dependencies and supply chain hygiene
Dependency information is incomplete, so supply-chain confidence stays limited.
What drove the risk score up
声明exec权限但无实际代码 +15
SKILL.md声明了exec/write工具但没有scripts目录，权限声明宽泛
功能描述模糊 +15
"自动化执行""命令执行""API集成"等描述过于宽泛，缺乏具体边界
依赖外部服务 +10
依赖grok.com服务，需浏览器会话
文档欺骗风险 +5
功能描述覆盖多个宽泛场景关键词
Most important evidence
Medium Doc Mismatch
权限声明宽泛且无实际代码支撑
Skill声明了read/exec/write工具权限，但没有scripts目录或实际可执行代码。文档中的Python/bash代码仅为示例，实际功能由AI Agent根据文档指令执行，存在"声明-执行"不一致风险。
SKILL.md:5 如无实际脚本，应移除exec/write权限声明；或提供实际脚本文件供审查
Medium Priv Escalation
功能描述过于宽泛
"自动化执行""命令执行""在安全沙箱中执行系统命令"等描述未限定具体操作范围，可能超出用户预期。
SKILL.md:410 明确限定可执行的命令类型和范围
Low Supply Chain
依赖外部服务grok.com
功能依赖Grok Imagine服务，需已登录的浏览器会话，存在服务可用性和数据隐私风险。
SKILL.md:140 评估对外部服务的依赖风险
Declared capability vs actual capability
Filesystem Pass
Declared WRITE
→ Inferred NONE
SKILL.md 声明write权限但无实际文件写入代码 Shell Pass
Declared WRITE
→ Inferred NONE
SKILL.md 声明exec权限但无实际脚本，仅有示例代码 Network Pass
Declared NONE
→ Inferred READ
SKILL.md:140 引用grok.com/imagine Browser Pass
Declared NONE
→ Inferred WRITE
SKILL.md 描述浏览器自动化功能（Playwright） Suspicious artifacts and egress
Medium External URL
https://grok.com/imagine SKILL.md:140
Dependencies and supply chain
There are no structured dependency warnings.
File composition
1 files · 502 lines
Markdown 1 files · 502 lines
Files of concern · 1
SKILL.md 权限声明宽泛且无实际代码支撑 · 功能描述过于宽泛 · 依赖外部服务grok.com · https://grok.com/imagine
Security positives
纯文档型Skill，无实际可执行代码，风险相对可控
文档包含安全注意事项章节
示例代码展示合理的工作流，无明显恶意模式
无敏感路径访问、无凭证收割、无数据外泄