🛡️ SecureShield AI Gateway

An enterprise-grade AI security middleware and proxy that acts as an intelligent "bouncer" for Large Language Models (LLMs), enforcing data governance, preventing prompt injection, and scrubbing PII in real-time.

🌟 Features

• ⚡ Multi-Layered Security Pipeline: Requests pass through Unicode Normalization, Input Guard, RBAC Policy Engine, Toxicity Guard, Semantic Guard, and PII Redaction.
• 🛡️ Active Rate Limiting (Network Guard): Middleware rate limiter to restrict request velocity and defend against DDoS attacks.
• 👤 Role-Based Access Control (RBAC): Fine-grained, department-level prompt safety policies (e.g., HR, Finance, and general employee constraints).
• 🤖 LLM-as-a-Judge Auditing: Concurrently evaluates prompt toxicity and semantic jailbreak attempts using high-speed cloud LLMs.
• 🔍 Advanced PII Redaction: Scrubs emails, phone numbers, API keys, and corporate vault secrets using spaCy, Microsoft Presidio, and custom regex pattern lists.
• 🔐 Secrets Leak Protection (Output Guard): Scans AI responses for leaked JWT tokens, API keys, and server credentials before they leave the gateway.
• 📊 Admin Monitoring Dashboard: Visualizes operational traffic, threat analytics, risk distribution graphs, attack correlation trends, and recent activity logs.

🛠️ Tech Stack

Frontend

• React.js & Vite - Modern, responsive web interface
• Tailwind CSS - Sleek, flexible, and premium dark-mode custom styles
• Recharts - Dynamic data visualization for traffic and threats
• Framer Motion - Fluid micro-animations and page transitions

Backend

• Python 3.12 - Core backend runtime
• FastAPI & Uvicorn - High-concurrency asynchronous endpoints
• MongoDB & Motor - Asynchronous database driver for logging, policies, and users

AI / Security / NLP

• Instructor - Strict JSON structural output enforcement
• OpenRouter API - Gateway connection to model providers (Llama 3.1 8b, Gemini 2.0 Flash)
• Microsoft Presidio & spaCy - Highly optimized NLP-based anonymization
• Lakera Guard API - Advanced third-party prompt injection analyzer

Project Structure

SecureShield-AI/
├── backend/             # FastAPI backend
│   ├── app/             # Server core directories
│   │   ├── layers/      # 6 pipeline security layers
│   │   ├── services/    # DB, users, logging, and AI handlers
│   │   └── config.py    # Environment settings
│   ├── tests/           # Pipeline test suites
│   ├── .env.example     # Configuration template
│   └── requirements.txt # Python dependencies
├── frontend/            # React + Vite frontend
│   ├── src/             # Frontend source files
│   │   ├── pages/       # Dashboard, chat, logs, settings
│   │   └── components/  # Charts, layout, and pipeline simulator
│   └── package.json     # Node.js dependencies
└── README.md            # Project documentation

⚙️ Installation

1️⃣ Clone the Repository

git clone <PRIVATE_URL>
cd SecureShield-AI

2️⃣ Setup Backend

cd backend

# Create virtual environment
python -m venv .venv

# Activate virtual environment
# Windows
.venv\Scripts\activate
# Linux/Mac
source .venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Start FastAPI server
python -m app.main

Backend runs on: http://localhost:8000

3️⃣ Setup Frontend

cd ../frontend

# Install dependencies
npm install

# Start Vite dev server
npm run dev

Frontend runs on: http://localhost:5173

🔥 How It Works

1. Submit Prompt: A user submits a query via the chat dashboard.
2. Standardize & Intercept: Network rate limits apply and Unicode normalizes characters to prevent bypasses.
3. Policy & Intent Scan: Role-based access controls and concurrent LLM judges scan for malicious prompt injection or toxic content.
4. PII Scrubbing: SpaCy, Presidio, and the corporate vault redact emails, phone numbers, and keys.
5. Secure Dispatch: The clean, safe prompt is forwarded to the backend LLM brain.
6. Response Audit: The response is validated to block or scrub secret leakage before hitting the user interface.

🚀 Future Enhancements

• Active Firewall Blocking: Real-time IP banning for users triggering recurrent high-severity security threats.
• Exportable PDF Reports: Automated SOC2/GDPR compliance reporting.
• Unified SIEM Integrations: Export security telemetry straight to Splunk or Datadog.
• Multi-Model Routing: Intelligently dispatch safe prompts to different models depending on tasks and cost optimization.

🤝 Contributing Contributions are welcome!

1. Fork the repository
2. Create a feature branch (git checkout -b feature/NewFeature)
3. Commit your changes (git commit -m 'Add NewFeature')
4. Push to the branch (git push origin feature/NewFeature)
5. Open a Pull Request