🛡️ Veil Armor - LLM Security Framework

Veil Armor is an enterprise-grade security framework for Large Language Models (LLMs) that provides multi-layered protection against prompt injections, jailbreaks, PII leakage, and sophisticated attack vectors.

🎯 Key Features

• 100% Attack Detection Rate - Tested against 42 zero-day attack vectors
• Prompt Injection Detection - Real-time detection using Vigil TransformerScanner
• Jailbreak Prevention - 30+ custom regex patterns for bypasses Vigil misses
• PII Protection - Microsoft Presidio integration for sensitive data detection
• Real-time Security API - FastAPI-powered RESTful endpoints
• Kubernetes Ready - Health checks, metrics, and deployment manifests included
• Docker Support - Multi-stage production builds

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                     Veil Armor API                          │
├─────────────────────────────────────────────────────────────┤
│  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐      │
│  │   Vigil      │  │  Presidio    │  │   Custom     │      │
│  │  Scanner     │  │  PII Engine  │  │  Patterns    │      │
│  └──────────────┘  └──────────────┘  └──────────────┘      │
├─────────────────────────────────────────────────────────────┤
│                    FastAPI Server                           │
└─────────────────────────────────────────────────────────────┘

📋 Requirements

• Python 3.10+
• CUDA (optional, for GPU acceleration)
• Docker (optional, for containerized deployment)

🚀 Quick Start

1. Clone and Setup

git clone https://github.com/0x-Professor/VeilArmor.git
cd veil-armor
python -m venv venv
venv\Scripts\activate  # Windows
source venv/bin/activate  # Linux/Mac
pip install -r requirements.txt
python -m spacy download en_core_web_lg

2. Configure Environment

Create a .env file:

VEIL_ARMOR_API_KEY=your_secret_api_key_here
GEMINI_API_KEY=your_gemini_key_here  # Optional
HF_TOKEN=your_huggingface_token_here  # For chatbot models

3. Start the Security API

cd src/veil_armor/api
python server.py

The API will be available at http://localhost:8000

🔐 API Usage

Health Check

curl http://localhost:8000/health

Security Check

curl -X POST http://localhost:8000/api/v1/check \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your_api_key" \
  -d '{
    "prompt": "Your user input here",
    "user_id": "user123",
    "check_pii": true,
    "check_injection": true
  }'

Response Format

{
  "safe": true,
  "threats_detected": [],
  "risk_score": 0.0,
  "pii_detected": null,
  "sanitized_prompt": null,
  "processing_time_ms": 45.23,
  "request_id": "req_1234567890"
}

🔍 Detection Capabilities

Prompt Injection Detection

• Vigil TransformerScanner (protectai/deberta-v3-base-prompt-injection)
• Confidence threshold: 0.8

Jailbreak Pattern Detection

• Developer/Admin mode bypasses
• AIM/Machiavellian persona attacks
• Hypothetical/fictional scenario attacks
• Grandma/emotional manipulation exploits
• Translation bypass attempts
• Context manipulation attacks
• Function/tool call injections
• Authority claim impersonation
• Code execution attempts

PII Detection (Presidio)

• Email addresses
• Phone numbers
• Credit card numbers
• Social Security Numbers (SSN)
• Passport numbers
• IP addresses
• Bank account numbers
• IBAN codes
• Driver's license numbers
• Cryptocurrency addresses

🐳 Docker Deployment

Build and Run API

# Build
docker build -t veil-armor:latest .

# Run
docker run -d \
  --name veil-armor \
  -p 8000:8000 \
  -e VEIL_ARMOR_API_KEY=your_key \
  veil-armor:latest

Docker Compose

docker-compose up -d

🤖 Chatbot Integration

Veil Armor includes a secure chatbot demo:

cd chatbot
pip install -r requirements.txt

# Run secure version (with Veil Armor protection)
streamlit run app_secure.py

# Run unsecure version (for comparison)
streamlit run app_unsecure.py

📊 API Endpoints

Endpoint	Method	Description
/	GET	Service info
/health	GET	Health check
/ready	GET	Readiness probe
/metrics	GET	Prometheus metrics
/api/v1/check	POST	Security analysis
/api/v1/generate	POST	Secure LLM generation
/api/v1/stats	GET	Real-time statistics

⚙️ Configuration

Environment Variables

Variable	Description	Default
VEIL_ARMOR_API_KEY	API authentication key	veil_armor_secret_key_12345
VEIL_ARMOR_API_URL	API base URL	http://localhost:8000
GEMINI_API_KEY	Google Gemini API key	-
HF_TOKEN	Hugging Face token	-

📁 Project Structure

veil-armor/
├── src/
│   └── veil_armor/
│       ├── api/
│       │   └── server.py      # Main API server
│       ├── middleware/        # Security middleware
│       ├── scanners/          # Detection modules
│       ├── security/          # Enterprise security
│       └── utils/             # Utilities
├── chatbot/
│   ├── app_secure.py          # Secured chatbot
│   ├── app_unsecure.py        # Unsecured chatbot
│   └── security_client.py     # API client
├── tests/
│   └── test_zero_day_attacks.py  # Attack test suite
├── kubernetes/
│   └── deployment.yaml        # K8s manifests
├── Dockerfile
├── docker-compose.yml
├── requirements.txt
└── README.md

🧪 Testing

Run the security test suite:

cd tests
pytest test_zero_day_attacks.py -v

Expected: 42/42 tests passing (100% detection rate)

📈 Metrics

Access Prometheus-compatible metrics at /metrics:

veil_armor_requests_total
veil_armor_requests_blocked
veil_armor_requests_allowed
veil_armor_uptime_seconds

🔒 Security Best Practices

1. Always use HTTPS in production
2. Rotate API keys regularly
3. Enable rate limiting for public endpoints
4. Monitor blocked requests for attack patterns
5. Keep dependencies updated

📄 License

Apache 2.0

🤝 Support

For enterprise support and custom implementations, contact the development team.

---

Veil Armor - Protecting your LLM applications from sophisticated attacks.