Documentation

🚀 Quick Start Guide

Get up and running with perfecX Red-T in just 10 minutes! This guide will walk you through setting up your first AI red team testing environment.

Prerequisites

System Requirements

Minimum

4 CPU cores
8GB RAM
20GB storage
Docker 20.10+

1.Installation

Docker Installation (Recommended)

# Pull the Red-T container
docker pull perfecxion/red-t:latest

# Create configuration directory
mkdir -p ~/.perfecx-red-t/config

# Run Red-T with default configuration
docker run -d \
  --name perfecx-red-t \
  -p 8080:8080 \
  -p 8443:8443 \
  -v ~/.perfecx-red-t/config:/app/config \
  -e REDTEAM_LICENSE_KEY="your-license-key" \
  perfecxion/red-t:latest

# Verify installation
docker ps | grep perfecx-red-t

Alternative: Standalone Installation

# Download Red-T CLI
curl -L https://releases.perfecxion.ai/red-t/latest/red-t-linux-amd64 -o red-t
chmod +x red-t
sudo mv red-t /usr/local/bin/

# Initialize configuration
red-t init --license-key="your-license-key"

# Start Red-T server
red-t server --port 8080

2.Initial Configuration

Basic Configuration

Create your first configuration file to define your target AI system:

# ~/.perfecx-red-t/config/targets.yaml
targets:
  - name: "production-chatbot"
    type: "llm_api"
    endpoint: "https://api.yourcompany.com/chat"
    auth:
      type: "bearer"
      token: "${API_TOKEN}"
    
    # Test configuration
    test_config:
      max_concurrent: 5
      timeout: 30
      retry_attempts: 3
    
    # Scope definition
    scope:
      attack_types:
        - "prompt_injection"
        - "model_inversion"
        - "data_poisoning"
        - "adversarial_examples"
      
      risk_levels: ["low", "medium", "high"]
      
    # Safety limits
    safety:
      max_requests_per_minute: 100
      stop_on_critical: true
      preserve_production: true

Environment Setup

# .env file
REDTEAM_LICENSE_KEY=your-license-key-here
API_TOKEN=your-target-api-token

# Database configuration
POSTGRES_HOST=localhost
POSTGRES_DB=redteam_db
POSTGRES_USER=redteam
POSTGRES_PASSWORD=secure-password

# Security settings
JWT_SECRET=your-jwt-secret
ENCRYPTION_KEY=your-32-char-encryption-key

# Optional: AI model configuration
OPENAI_API_KEY=your-openai-key  # For enhanced attack generation
HUGGINGFACE_TOKEN=your-hf-token  # For local model testing

3.Your First Red Team Test

Web Interface

Access the Dashboard: Open http://localhost:8080 in your browser

Login: Use default credentials (admin/admin) or your configured credentials
Create Target: Navigate to "Targets" → "Add New Target"
Configure Test: Select "Quick Test" from the dashboard
Choose Attack Types: Select "Prompt Injection" for your first test
Run Test: Click "Execute Test" and monitor progress

CLI Interface

# Quick vulnerability scan
red-t scan --target production-chatbot --type prompt-injection

# Run comprehensive assessment
red-t assess --target production-chatbot --full-suite

# Generate test report
red-t report --scan-id scan_123456 --format pdf

# Real-time monitoring
red-t monitor --target production-chatbot --live

API Testing

# Test via API
curl -X POST http://localhost:8080/api/v1/scans \
  -H "Authorization: Bearer your-api-token" \
  -H "Content-Type: application/json" \
  -d '{
    "target_id": "production-chatbot",
    "attack_types": ["prompt_injection"],
    "intensity": "medium",
    "max_duration": 300
  }'

# Check scan status
curl -X GET http://localhost:8080/api/v1/scans/scan_123456 \
  -H "Authorization: Bearer your-api-token"

# Get results
curl -X GET http://localhost:8080/api/v1/scans/scan_123456/results \
  -H "Authorization: Bearer your-api-token"

4.📊 Understanding Your Results

Risk Assessment Score

0-25

Low Risk

26-50

Medium Risk

51-75

High Risk

76-100

Critical Risk

Common Findings

Critical: Prompt Injection Successful

The AI system accepted malicious prompts that could bypass safety filters or extract sensitive information.

High: Model Inversion Detected

Potential training data extraction through carefully crafted queries.

Medium: Rate Limiting Issues

Insufficient rate limiting could enable abuse or denial-of-service attacks.

5.🎯 Next Steps

Immediate Actions

• Review and address critical findings
• Implement recommended security controls
• Set up automated testing schedules
• Configure alerting for new vulnerabilities

Advanced Configuration

• Custom attack scenarios
• API integration
• CI/CD pipeline integration
• Team collaboration setup

Sample Test Output

perfecX Red-T - AI Security Assessment Report
===========================================

Target: production-chatbot
Scan ID: scan_20240115_001
Duration: 5m 34s
Tests Executed: 47

RISK ASSESSMENT SUMMARY
┌─────────────────┬───────┬─────────┐
│ Risk Level      │ Count │ Percent │
├─────────────────┼───────┼─────────┤
│ Critical        │   2   │   4.3%  │
│ High           │   5   │  10.6%  │
│ Medium         │  12   │  25.5%  │
│ Low            │  28   │  59.6%  │
└─────────────────┴───────┴─────────┘

VULNERABILITY BREAKDOWN
• Prompt Injection: 3 successful attempts
• Model Inversion: 2 potential data leaks  
• Rate Limiting: Bypassed in 4/5 attempts
• Input Validation: 8 filter bypasses
• Authentication: No issues detected

RECOMMENDATIONS
1. Implement robust input sanitization
2. Add rate limiting per user/IP
3. Enhance prompt filtering rules
4. Monitor for anomalous query patterns
5. Regular security assessments

Full report available at: /reports/scan_20240115_001.pdf

Important Security Notice

Only test systems you own or have explicit permission to test. Red team testing can generate significant load and potentially disruptive traffic. Always coordinate with system owners and follow responsible disclosure practices.

Continue Your Journey

📦 Complete Installation Guide →🔌 API Reference Documentation →📚 Advanced Guides & Examples →