Aura: AI-Powered Web Accessibility Auditor 🚀

Aura is a full-stack web accessibility platform that combines Axe-core scanning with AI-powered suggestions to achieve WCAG compliance.

🤖 AI Features

1. BLIP Image Captions - Generates alt text using Hugging Face's BLIP model
2. Color Contrast AI - Calculates WCAG-compliant color fixes (4.5:1 ratio)
3. Flesch-Kincaid Analysis - Analyzes text readability grade level

🛠️ Tech Stack

Backend: Python, FastAPI, MongoDB, Playwright, Axe-core
AI/ML: Hugging Face Transformers (BLIP), PyTorch, TextStat
Frontend: HTML, CSS, JavaScript

📦 Installation

Prerequisites

• Python 3.11+
• MongoDB Atlas account

Setup

1. Clone & Install

git clone https://github.com/architzero/Aura-accessibility-scanner.git
cd Aura-accessibility-scanner/backend
python -m venv venv
venv\Scripts\activate  # Windows
pip install -r requirements.txt
playwright install chromium

2. Configure Environment

Create backend/.env:

MONGO_DB_URI="your-mongodb-connection-string"
JWT_SECRET_KEY="your-secret-key"  # Generate: python -c "import secrets; print(secrets.token_hex(32))"
JWT_ALGORITHM="HS256"
CORS_ORIGINS=["http://127.0.0.1:8001","http://localhost:8001"]
ENVIRONMENT="development"

3. Run Application

# Terminal 1 - Backend
cd backend
uvicorn main:app --reload

# Terminal 2 - Frontend
cd frontend
python -m http.server 8001

4. Access: http://127.0.0.1:8001

🧪 Test AI Features

cd backend
python test_ai_features.py

Expected output:

• ✅ BLIP Model Working!
• ✅ Color Contrast AI Working!
• ✅ Flesch-Kincaid Working!

🌐 Test Sites

Recommended:

• https://example.com - Simple baseline
• https://www.berkshirehathaway.com - Contrast issues
• https://www.arngren.net - Multiple issues
• https://news.ycombinator.com - Fast scan

Blocked (CSP):

• Instagram, Facebook, Twitter, LinkedIn

📊 How It Works

1. User creates project with URL
2. Playwright + Axe-core scan website
3. AI analyzes violations:
   • Image-alt → BLIP generates descriptions
   • Color-contrast → Calculates WCAG fixes
   • Text → Flesch-Kincaid readability
4. Results stored with severity-based score
5. User views issues + AI suggestions

🎯 Key Features

• ✅ Live website audits
• ✅ WCAG 2.1 compliance checking
• ✅ AI-powered fix suggestions
• ✅ Severity-based scoring (Critical: 10pts, Serious: 7pts, Moderate: 4pts, Minor: 2pts)
• ✅ Full-page screenshots
• ✅ Scan history tracking
• ✅ JWT authentication
• ✅ Responsive UI

📁 Project Structure

backend/
  scanner_process.py    # Main AI scanner
  routers/              # API endpoints
  services/             # Business logic
  config.py             # Configuration
  
frontend/
  dashboard.html        # Projects & scanning
  results.html          # AI suggestions
  style.css             # Modern UI

🔒 Security

• JWT authentication
• Password hashing (bcrypt)
• URL validation (SSRF prevention)
• CORS protection
• Input sanitization

📝 License

MIT License

👨‍💻 Author

Archit - @architzero

---

Built with ❤️ for a more accessible web