📘 SafeguardAI Quick Reference

A quick reference guide for common SafeguardAI operations.

Installation

npm install safeguard-ai

Basic Setup

const SafeguardAI = require('safeguard-ai');
require('dotenv').config();

const moderator = new SafeguardAI({
    apiKey: process.env.OPENAI_API_KEY,
    strictness: 'medium',  // 'low' | 'medium' | 'high'
    redactPII: true
});

Core Methods

Check Text

const result = await moderator.checkText("Your text here");

// Result object:
{
    safe: boolean,           // Is content safe?
    flagged: boolean,        // Was content flagged?
    categories: {...},       // Violation categories
    piiDetected: [...],      // PII found
    cleanText: string,       // Redacted text
    suggestions: [...]       // Recommendations
}

PII Detection

Detect PII

const piiItems = moderator.detector.detect("Email: john@example.com");
// Returns array of PII objects

Redact PII

// Mode: 'replace' | 'mask' | 'fake'
const result = moderator.detector.redact("john@example.com", 'mask');
console.log(result.redactedText); // joh***@example.com

Custom Rules

Add Blocked Words

moderator.rules.addBlockedWords(['spam', 'scam', 'phishing']);

Add Custom Patterns

// Detect API keys
moderator.rules.addPattern(/API-KEY-\w{20}/g, 'API_KEY');

// Detect employee IDs
moderator.rules.addPattern(/EMP-\d{6}/g, 'EMPLOYEE_ID');

// Detect SSN
moderator.rules.addPattern(/\d{3}-\d{2}-\d{4}/g, 'SSN');

Configuration Options

new SafeguardAI({
    apiKey: string,              // Required: OpenAI API key
    providers: ['openai'],       // Providers to use
    strictness: 'medium',        // 'low' | 'medium' | 'high'
    categories: string[],        // Specific categories to check
    redactPII: boolean           // Auto-redact PII
})

Common Patterns

Basic Validation

async function validateContent(text) {
    const result = await moderator.checkText(text);
    return result.safe;
}

Express Middleware

async function moderationMiddleware(req, res, next) {
    const result = await moderator.checkText(req.body.content);
    
    if (!result.safe) {
        return res.status(400).json({ error: 'Content flagged' });
    }
    
    req.body.cleanContent = result.cleanText;
    next();
}

Chatbot Filter

async function filterMessage(message) {
    const result = await moderator.checkText(message);
    
    if (!result.safe) {
        return "I cannot process inappropriate content.";
    }
    
    return result.cleanText;
}

Error Handling

try {
    const result = await moderator.checkText(text);
} catch (error) {
    if (error.response?.status === 429) {
        // Rate limit error
    } else if (error.response?.status === 401) {
        // Invalid API key
    } else {
        // Other error
    }
}

Environment Setup

Create .env file:

OPENAI_API_KEY=sk-your-api-key-here

Load in your app:

require('dotenv').config();

PII Types Detected

Strictness Levels

Response Categories

Moderation categories checked:

• toxicity - Toxic/offensive content
• hate - Hate speech
• violence - Violent content
• sexual - Sexual content
• self-harm - Self-harm content
• harassment - Harassment
• custom - Custom rules violations

Testing

// Quick test script
async function test() {
    console.log('Testing SafeguardAI...');
    
    const tests = [
        { text: "Hello world", expected: true },
        { text: "Email: test@test.com", expected: true }
    ];
    
    for (const test of tests) {
        const result = await moderator.checkText(test.text);
        console.log(`${test.text}: ${result.safe ? '✅' : '❌'}`);
    }
}

Performance Tips

1. Cache Results

const cache = new Map();

async function checkWithCache(text) {
    if (cache.has(text)) return cache.get(text);
    const result = await moderator.checkText(text);
    cache.set(text, result);
    return result;
}

2. Batch Processing

const results = await Promise.all(
    texts.map(text => moderator.checkText(text))
);

3. Rate Limiting

const rateLimit = require('express-rate-limit');

const limiter = rateLimit({
    windowMs: 15 * 60 * 1000, // 15 minutes
    max: 100 // limit each IP to 100 requests per windowMs
});

app.use('/api/', limiter);

Common Issues

"API key required"

// Make sure .env is loaded
require('dotenv').config();

"Module not found"

npm install safeguard-ai

Rate limit errors

// Add retry logic with exponential backoff

Slow responses

// Implement caching or async processing

Links

• Full Documentation: GETTING_STARTED.md
• API Reference: API.md
• Examples: ../examples/
• GitHub: ai-safety-toolkit
• NPM: safeguard-ai

Need Help?

1. Check GETTING_STARTED.md for detailed guide
2. Review examples/ for code samples
3. Open an issue on GitHub

Print this page for quick reference! 📄