低风险 — 风险评分 20/100
上次扫描:1 天前 重新扫描
20 /100
claim-risk-auditor
检查文案、论文、宣传稿或产品说明中的高风险断言,标出证据缺口并给出更稳妥的改写
A straightforward clipboard-reading utility for macOS that accurately declares its shell:WRITE permission and performs only read-only pbpaste execution with no hidden behavior.
技能名称claim-risk-auditor
分析耗时25.3s
引擎pi
可以安装
No action required. The skill is straightforward and its declared shell:WRITE permission is necessary for reading clipboard via pbpaste. Consider pinning node dependency and restricting permission to shell:READ if the tool allows it.

安全发现 2 项

严重性 安全发现 位置
低危
Clipboard read not declared in metadata 权限提升
The script reads clipboard content via pbpaste but SKILL.md metadata does not declare clipboard:READ as a required permission. This is a minor omission as the purpose is transparent and the behavior is documented in the script body.
return execSync('pbpaste', { encoding: 'utf8' }).trim();
→ Add clipboard:READ to the allowed-tools mapping in the capability model or document it in SKILL.md metadata.
scripts/read_clipboard.mjs:6
低危
shell:WRITE permission is inflated 权限提升
The metadata declares shell:WRITE but the script only executes pbpaste, a read-only command. A read-only shell or exec permission would be sufficient.
execSync('pbpaste', ...)
→ If the tool framework supports shell:READ, use that instead to accurately reflect the minimal privilege principle.
scripts/read_clipboard.mjs:6
资源类型声明权限推断权限状态证据
文件系统 NONE NONE No file reads or writes found
网络访问 NONE NONE No network requests found
命令执行 WRITE READ (read-only pbpaste) ✓ 一致 scripts/read_clipboard.mjs:6 — execSync runs 'pbpaste' which only reads clipboar…
环境变量 NONE NONE No environment variable access
剪贴板 NONE READ ✓ 一致 scripts/read_clipboard.mjs:6 — reads clipboard via pbpaste; clipboard:READ not d…

目录结构

4 文件 · 1.7 KB · 84 行
Markdown 3f · 65L JavaScript 1f · 19L
├─ 📁 scripts
│ └─ 📜 read_clipboard.mjs JavaScript 19L · 445 B
├─ 📝 CHANGELOG.md Markdown 6L · 114 B
├─ 📝 README.md Markdown 26L · 379 B
└─ 📝 SKILL.md Markdown 33L · 771 B

安全亮点

✓ No network requests or data exfiltration
✓ No credential harvesting or environment variable access
✓ No obfuscation, base64, or anti-analysis patterns
✓ No sensitive file or path access (~/.ssh, ~/.aws, .env)
✓ Script is short (19 lines) and easy to audit
✓ No external dependencies or npm packages
✓ Script behavior fully aligns with SKILL.md description
✓ No hidden functionality beyond stated purpose
✓ Includes proper error handling for empty clipboard