Framework-RAI: Responsible AI in Minutes

A lightweight, developer-friendly framework that ensures responsible AI practices with minimal effort. Generate essential documentation, receive actionable tips, and comply with ethical guidelines—all in under 3 minutes per feature.

🚀 Features

• Fast & Frictionless: Complete all documentation in under 3 minutes
• AI-Powered Tips: Receive 2-3 practical, actionable tips per category
• Minimal Questions: Answer only the most impactful questions
• Pre-commit Integration: Automate checks before code is committed
• Automatic Code Scanning: Detect AI libraries and functions in your codebase
• GitHub Actions Integration: Run framework-rai checks in your CI pipeline
• Comprehensive Documentation: Generate three essential documents:
  • checklist.md: Essential questions for every AI feature
  • model_card.md: Simple model documentation
  • risk_file.md: Minimal audit & compliance steps

📋 Why Use Framework-RAI?

Responsible AI shouldn't be a burden. This framework helps you:

• Save Time: No lengthy forms or complex processes
• Ensure Compliance: Cover essential ethical and regulatory requirements
• Improve Documentation: Create clear, consistent documentation
• Get Actionable Advice: Receive practical tips tailored to your AI feature

🔧 Installation

# Install globally
npm install -g framework-rai

# Or use without installing
npx framework-rai

Setting Up Your OpenAI API Key

The first time you run the tool, it will prompt you to set up your OpenAI API key. You can also configure it explicitly:

# Initial setup wizard (interactive)
npx framework-rai --setup

# Set up a global key (used across all projects)
npx framework-rai --setup --global

# Run with a specific key without saving it
npx framework-rai --key=your-api-key-here

API Key Storage Options

• Project-specific: Stored in .env in your project directory (default)
• Global: Stored in ~/.config/framework-rai/config (works across all projects)
• Session-only: Use the --key option for temporary use without saving

📝 Usage

Basic Usage

Run the CLI tool in your project directory:

# If installed globally
ai-responsible

# Or using npx
npx framework-rai

Answer the simple questions about your AI feature, and the framework will generate three markdown files:

• checklist.md
• model_card.md
• risk_file.md

CLI Options

# Basic usage
npx framework-rai

# Configure API key (interactive)
npx framework-rai --setup

# Use a global API key (shared between projects)
npx framework-rai --global

# Specify an API key directly (won't be saved)
npx framework-rai --key=your-api-key-here

# Scan codebase for AI-related code
npx framework-rai --scan

# Run in CI mode (non-interactive)
npx framework-rai --scan --ci

# Combine options
npx framework-rai --global --setup

Pre-commit Hook (Recommended)

For automatic checks before each commit, install Husky:

# Install Husky
npm install husky --save-dev
npx husky init

# Add the pre-commit hook
npx husky add .husky/pre-commit "npx framework-rai"

GitHub Actions Integration

You can automatically run Framework-RAI checks in your CI pipeline:

1. Create a GitHub Secret called OPENAI_API_KEY with your OpenAI API key
2. Add the workflow file to your repository:

# .github/workflows/rai-check.yml
name: Responsible AI Framework Check

on:
  push:
    branches: [ main, master ]
  pull_request:
    branches: [ main, master ]

jobs:
  rai-check:
    runs-on: ubuntu-latest
    
    steps:
    - uses: actions/checkout@v3
    - name: Set up Node.js
      uses: actions/setup-node@v3
      with:
        node-version: '16.x'
        cache: 'npm'
    
    - name: Install dependencies
      run: npm ci
    
    - name: Run Framework-RAI
      run: |
        node ./ai-responsible.mjs --scan --ci
      env:
        OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
    
    - name: Upload RAI Documentation
      uses: actions/upload-artifact@v3
      with:
        name: rai-documentation
        path: |
          checklist.md
          model_card.md
          risk_file.md

🔍 How It Works

1. Detection: Identifies if a commit introduces or modifies an AI feature
2. Code Scanning: Automatically detects AI libraries and functions
3. Documentation: Guides you through minimal, high-impact questions
4. Generation: Creates three essential markdown files
5. AI Tips: Provides 2-3 practical tips per category using OpenAI's API
6. Integration: Works with your existing workflow via pre-commit hooks, CI pipelines, or manual runs

Automatic Code Scanning

The code scanning feature can detect:

• AI Libraries: TensorFlow, PyTorch, scikit-learn, Hugging Face, etc.
• AI Functions: fit, predict, transform, evaluate, etc.
• Risk Patterns: Common patterns that may require additional documentation

When running with the --scan option, the tool will:

1. Recursively scan your codebase for AI-related patterns
2. Generate documentation templates based on the detected patterns
3. Provide the option to refine the templates with interactive mode

🧠 Example Output

checklist.md

# Responsible AI Feature Checklist

This AI feature is mainly for internal analytics, using personal data with consent. 
Monitoring approach: user feedback.

## Tips
Checklist Tips:
- Implement a consent management system to track and verify user consent for personal data.
- Document your user feedback collection process in the README.

model_card.md

# Model Card

Model type: NLP/Language. Key metric: F1 Score. Main limitation: Data bias.

## Tips
Model Card Tips:
- Include a section on data bias in your model card, detailing specific biases identified.
- Add F1 score benchmarks for different demographic groups to identify potential disparities.

📊 Metrics & KPIs

The framework tracks three key ethical KPIs:

1. Model Transparency: Presence of model documentation
2. Bias & Fairness: Consideration of potential biases
3. Monitoring Setup: Implementation of feedback mechanisms

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

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

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

Built with ❤️ for responsible AI development.