A hybrid AI-powered security auditing tool for scanning skill directories and generating visual security dashboards.
一款結合 AI 智慧分析的混合式安全審查工具,用於掃描技能目錄並生成視覺化安全儀表板。
Skills-Security-Check is a security scanning tool designed for AI Agent skill repositories. It combines:
Skills-Security-Check 是一款專為 AI Agent 技能倉庫設計的安全掃描工具,結合了:
- Static Analysis | 靜態分析 - Regex-based pattern matching to identify potential risks | 使用正則表達式匹配潛在風險
- AI Intelligence | AI 智慧分析 - Leverages AI agents to analyze findings and reduce false positives | 利用 AI 代理分析發現並減少誤報
- Visual Dashboard | 視覺化儀表板 - Generates a beautiful, interactive HTML dashboard | 生成精美的互動式 HTML 儀表板
| Category | 類別 | Examples | 範例 |
|---|---|---|---|
| 🔑 Sensitive Operations | 敏感操作 | API keys, credentials, environment variables | API 金鑰、憑證、環境變數 |
| 🌐 Network Activity | 網路活動 | External URLs, IP addresses, API endpoints | 外部連結、IP 位址、API 端點 |
| 🎭 Obfuscation Signals | 混淆跡象 | Base64 encoding, eval(), dynamic imports | Base64 編碼、eval()、動態載入 |
| 📦 Package Installs | 套件安裝 | npm, pip, apt, brew, yarn, pnpm, gem, go | npm, pip, apt, brew 等安裝指令 |
| 高風險模式 | Shell execution, download-and-execute | Shell 執行、下載並執行 |
- Python 3.8+
- No external dependencies required (uses standard library only)
- 無需外部依賴(僅使用 Python 標準函式庫)
# Clone the repository | 複製專案
git clone https://github.com/YOUR_USERNAME/Skills-Security-Check.git
# Navigate to the skill directory | 進入技能目錄
cd Skills-Security-Check# Scan a directory of skills | 掃描技能目錄
python3 scripts/scan_skills.py --root /path/to/your/skills
# The dashboard will auto-open in your browser
# 儀表板將自動在瀏覽器中開啟reports/YYYYMMDD_HHMMSS/
├── index.html # Interactive dashboard | 互動式儀表板
├── data.json # Raw scan data | 原始掃描資料
└── prompts/ # AI audit prompts | AI 審查提示詞
├── skill1_audit_prompt.txt
└── skill2_audit_prompt.txt
This skill is designed to work with AI agents. The recommended workflow:
此技能專為 AI 代理設計,建議的工作流程如下:
- Run Scanner | 執行掃描 → Generates raw findings and audit prompts | 生成原始發現與審查提示詞
- AI Analysis | AI 分析 → Agent reads prompts and creates
audit.jsonfor each skill | 代理讀取提示詞並為每個技能建立audit.json - Integrate & Present | 整合呈現 → Re-run scanner to merge AI insights into final report | 重新執行掃描器以合併 AI 洞察至最終報告
See SKILL.md for detailed agent instructions.
詳細的代理指示請參閱 SKILL.md。
- Executive Summary | 總覽摘要 - Overall security score and top risks at a glance | 一目了然的安全評分與高風險項目
- Risk Filtering | 風險篩選 - Filter by High/Medium/Low risk levels | 依高/中/低風險等級篩選
- Detailed Views | 詳細檢視 - Click any skill to see full breakdown | 點擊任何技能查看完整分析
- AI Insights Card | AI 洞察卡片 - Displays AI-generated analysis when available | 顯示 AI 生成的分析結果
- Responsive Design | 響應式設計 - Works on desktop and tablet | 支援桌面與平板裝置
| Argument | 參數 | Description | 說明 | Default | 預設值 |
|---|---|---|---|---|---|
--root |
Root directory containing skills to scan | 包含待掃描技能的根目錄 | Current directory | 當前目錄 | |
--out |
Custom output path for HTML report | 自訂 HTML 報告輸出路徑 | Auto-generated | 自動生成 |
Skills-Security-Check/
├── SKILL.md # AI agent instructions | AI 代理指示
├── README.md # This file | 本檔案
├── scripts/
│ └── scan_skills.py # Main scanner script | 主掃描腳本
├── assets/
│ └── dashboard_template.html # Dashboard HTML template | 儀表板 HTML 模板
└── reports/ # Generated reports | 生成的報告 (gitignored)
Contributions are welcome! Please feel free to submit a Pull Request.
歡迎貢獻!請隨時提交 Pull Request。
Prompt Case
- 🧵 Threads: @prompt_case
- 💖 Patreon: MattTrendsPromptEngineering
This project is licensed under the MIT License.
本專案採用 MIT 授權條款。
Built with ❤️ for the AI Agent ecosystem.
為 AI Agent 生態系統用心打造 ❤️
