中文 EN

扫描报告

扫描新文件

⚠ 低风险 — 风险评分 15/100
上次扫描：1 天前 重新扫描
15 /100
skill-composer
编排多个OpenClaw技能成自动化工作流，一次命令完成复杂任务
Skill Composer is a legitimate workflow orchestrator for OpenClaw skills with no malicious indicators; uses subprocess for documented skill invocation and restricted eval for condition evaluation.
技能名称skill-composer
分析耗时28.8s
引擎pi
✓
可以安装
No immediate action required. Consider pinning PyYAML version in install.sh for supply chain hygiene.

安全发现 2 项

严重性 安全发现 位置
低危
Unpinned PyYAML dependency 供应链
install.sh installs PyYAML without specifying a version, potentially pulling a compromised or incompatible future version
pip3 install --user PyYAML
→ Pin version: pip3 install --user 'PyYAML>=5.4,<7.0'
 install.sh:21
低危
Broad skill orchestration scope not prominently documented 文档欺骗
skill.json declares 'skills: ["*"]' allowing execution of any installed skill, which could enable chained attacks if one skill is compromised
"skills": ["*"]
→ Document the '*' scope in SKILL.md with security considerations
 skill.json:28
资源类型声明权限推断权限状态证据
命令执行 WRITE WRITE ✓ 一致 source/composer.py:95 - subprocess.run(['claw', 'skill', 'exec', skill] + args)
文件系统 NONE READ ✓ 一致 source/composer.py:51 - reads workflow YAML files
网络访问 NONE NONE — No network requests observed
1 项发现
🔗
中危 外部 URL 外部 URL
https://utopiabenben.github.io/ai-skills/
SKILL.md:207

目录结构

7 文件 · 19.1 KB · 703 行
Python 1f · 241L Markdown 1f · 206L YAML 3f · 140L JSON 1f · 83L Shell 1f · 33L
├─ ▾ 📁 examples
│ ├─ 📋 content-creation-pipeline.yaml YAML 51L · 1.2 KB
│ ├─ 📋 video-processing-pipeline.yaml YAML 39L · 998 B
│ └─ 📋 weekly-stock-report.yaml YAML 50L · 1.1 KB
├─ ▾ 📁 source
│ └─ 🐍 composer.py Python 241L · 8.1 KB
├─ 🔧 install.sh Shell 33L · 926 B
├─ 📋 skill.json JSON 83L · 2.0 KB
└─ 📝 SKILL.md Markdown 206L · 4.9 KB

依赖分析 1 项

包名版本来源已知漏洞备注
PyYAML >=5.4 pip 否 Version not pinned in install.sh, but skill.json specifies >=5.4

安全亮点

✓ Uses yaml.safe_load() - safe YAML parsing
✓ eval() has restricted builtins: {"__builtins__": {}}
✓ No credential harvesting or environment variable enumeration
✓ No base64 encoding/decoding or obfuscation
✓ No external IP connections or data exfiltration
✓ subprocess only invokes local 'claw' command for skill execution
✓ No sensitive file path access (~/.ssh, ~/.aws, .env)
✓ No reverse shell, C2, or remote code execution indicators