MoltGuard-Security
由 Gemini 2.5 Flash 驅動的行動安全編排器,用於 AI 代理,可即時偵測提示注入與資料外洩
- 星標
- 1
- 分支
- 0
- 最近推送(UTC)
- 2026年3月21日
- 專案狀態
- 未封存
原始 README
以下為英文專案原文快照,最新內容請造訪 GitHub。
🛡️ MoltGuard: AI Security Orchestrator
If you find this project useful, please consider giving it a ⭐ to support its development!
MoltGuard is a cutting-edge mobile security orchestrator powered by the next-generation Gemini 2.5 Flash. It performs high-speed behavioral auditing on AI interactions to detect prompt injections, malware patterns, and data leaks with unprecedented reasoning capabilities.
⚙️ Quick Setup & API Configuration
To get started with MoltGuard, follow these steps:
- Obtain API Key: Get your free API key from Google AI Studio.
- Environment Setup: Create a
.envfile in the root directory. - Add Key: Add your key to the file:
GEMINI_API_KEY=your_key_here. - Run: Execute
flutter runin your terminal.
🧠 How the Security Analysis Works
MoltGuard acts as a middleware between the user and the LLM.
- Input Sanitization: User prompts are intercepted.
- Behavioral Auditing: Gemini 2.5 Flash analyzes the intent (detecting Jailbreaks, PII Leaks, or Malicious Code).
- Structured Response: The AI returns a precise JSON object containing:
overallRiskScore: Numerical value (0-100).vulnerabilities: Detailed list of detected threats.
- Actionable Output: Users receive a visual risk level and a downloadable PDF report.
🚀 Key Features (Powered by Gemini 2.5 Flash)
- Extreme Reasoning Speed: Leveraging the ultra-fast Gemini 2.5 Flash for real-time threat analysis without latency.
- Lux Loading Radar: A premium security scanning animation for an immersive "Cybersecurity" experience.
- Smart Risk Classification: Internal logic that auto-calculates threat levels (Critical, High, Medium, Low) ensuring no "Unknown" results.
- Professional PDF Audits: Generate detailed security reports with QR Code verification, powered by deep AI insights.
- Modular Clean Architecture: Organized into a professional Service Layer (Models, Services, Screens) for enterprise scalability.
📸 Preview





MoltGuard in action: Real-time auditing and professional reporting powered by Gemini 2.5 Flash.
📂 Project Structure
lib/
├── models/ # Data Structures & Smart Fallback Logic
├── services/ # AI Service (Gemini 2.5), Export (PDF), & Storage
├── screens/ # Premium UI Components
└── main.dart # Application entry point
🗺️ Future Roadmap
We have ambitious plans for MoltGuard! From on-device ML to enterprise-grade security monitoring. 🚀 View our detailed Roadmap here
🛠️ Built With
- Flutter - High-performance core cross-platform framework.
- Gemini 2.5 Flash - Utilizing the next generation of AI for advanced safety auditing.
- FlutLab - Professional cloud IDE used for rapid development and orchestration.
- MIT License - Open for community contribution and safety research.
⚠️ Disclaimer
This tool is for educational and research purposes only. Always use AI responsibly and ensure you comply with data privacy regulations.
Note: This project showcases the cutting-edge application of Gemini 2.5 Flash in the field of AI safety and mobile security guardrails. It aims to empower users with a transparent layer of protection in an AI-driven world.
分類
使用的依賴
依賴清單 9 項
- http
^1.1.0 - flutter
{"sdk":"flutter"} - flutter_dotenv
^5.1.0 - google_generative_ai
^0.4.0 - shared_preferences
^2.2.2 - pdf
^3.10.8 - path_provider
^2.1.2 - flutter_test開發依賴
{"sdk":"flutter"} - flutter_lints開發依賴
^4.0.0
如何選擇適合專案的儲存庫?
先依使用情境篩選,再比較專案活躍情況,最後到原始儲存庫確認使用要求。
- 透過儲存庫名稱、作者或描述搜尋,再依技術分類縮小範圍。
- 綜合星標、分支與最近推送比較專案;星標反映社群關注,不保證品質。
- 使用前在 GitHub 確認 README、授權條款與 Issues。
來源與維護
“A repository is the most basic element of GitHub.”
GitHub Docs — About repositories ↗