中文 EN

扫描报告

扫描新文件

⚠ 低风险 — 风险评分 20/100
上次扫描：1 天前 重新扫描
20 /100
smart-model-switcher-pro
智能模型自动切换 Pro V6.0.0 - 多模态感知，自动识别图片/视频/音频/代码/文本任务
Legitimate model switching utility with undocumented file system access and missing allowed-tools declaration, but no malicious behavior detected.
技能名称smart-model-switcher-pro
分析耗时28.4s
引擎pi
✓
可以安装
Add explicit allowed-tools declaration in SKILL.md frontmatter listing filesystem:READ (for config path check) and filesystem:WRITE (for log creation). Document the $env:USERPROFILE\.openclaw\ paths used by the scripts.

安全发现 2 项

严重性 安全发现 位置
低危
Missing allowed-tools declaration 文档欺骗
SKILL.md frontmatter does not declare any allowed-tools permissions despite scripts accessing file system paths ($env:USERPROFILE\.openclaw\) for config reading and log writing.
---
name: smart-model-switcher-pro
...
→ Add 'allowed-tools: [Read, Write]' to frontmatter to declare filesystem:READ/WRITE permissions.
 SKILL.md:1
低危
Undocumented file path access 文档欺骗
Scripts reference $env:USERPROFILE\.openclaw\ paths which are not mentioned in SKILL.md documentation.
[string]$ConfigPath = "$env:USERPROFILE\.openclaw\openclaw.json"
→ Document the data directory usage in SKILL.md capabilities section.
 scripts/runtime-switch.txt:7
资源类型声明权限推断权限状态证据
文件系统 NONE READ ✓ 一致 runtime-switch.txt:7 - reads $env:USERPROFILE\.openclaw\openclaw.json
文件系统 NONE WRITE ✓ 一致 auto-monitor.txt:16 - writes to $env:USERPROFILE\.openclaw\logs\
网络访问 NONE NONE — No network calls in any script
命令执行 NONE NONE — No shell execution via subprocess/popen

目录结构

3 文件 · 13.8 KB · 385 行
Text 2f · 268L Markdown 1f · 117L
├─ ▾ 📁 scripts
│ ├─ 📄 auto-monitor.txt Text 90L · 3.4 KB
│ └─ 📄 runtime-switch.txt Text 178L · 6.4 KB
└─ 📝 SKILL.md Markdown 117L · 3.9 KB

安全亮点

✓ No base64-encoded payloads or eval() calls
✓ No curl|bash or wget|sh remote script execution
✓ No credential harvesting (no os.environ iteration for secrets)
✓ No sensitive path access (~/.ssh, ~/.aws, .env files)
✓ No external network requests or C2 communication
✓ No obfuscation techniques detected
✓ No malicious dependencies - no package manager files present
✓ PowerShell scripts contain straightforward, readable logic for model selection