Best Practices

Production-ready patterns for building with RegPilot.

Security

1. Protect API Keys

// ✅ Good - Environment variables
const apiKey = process.env.REGPILOT_API_KEY;

// ❌ Bad - Hardcoded
const apiKey = 'sk_abc123';

// ❌ Bad - Client-side
// Never expose in browser JavaScript!

2. Use Secure Headers

// ✅ Production-safe (won't be stripped)
headers: {
  'X-API-Key': process.env.REGPILOT_API_KEY
}

// ⚠️ May be stripped by proxies
headers: {
  'Authorization': `Bearer ${apiKey}`
}

3. Rotate Keys Regularly

# Rotate every 90 days
Create new key
Update all environments
Verify new key works
Revoke old key

Performance

1. Choose Right Quality Tier

// Simple FAQ - use cheap
await chat('What are your hours?', 'cheap');

// General use - balanced
await chat('Explain quantum computing', 'balanced');

// Complex tasks - frontier
await chat('Review this legal contract', 'frontier');

2. Leverage Caching

// Cache FAQ responses
const faqs = [
  'What are your hours?',
  'Where are you located?',
  'What is your return policy?'
];

// First time: Cache MISS
// Subsequent: Cache HIT (FREE!)
faqs.forEach(q => chat(q, 'cheap'));

3. Optimize Prompts

// ❌ Too verbose
const prompt = `I would like you to please analyze this text and provide me with a detailed summary of the main points, key takeaways, and any important insights that you can identify from the content...`;

// ✅ Concise
const prompt = `Summarize the main points:`;

4. Manage Context Windows

// Keep under 4000 tokens
function trimHistory(messages: Message[]) {
  const maxTokens = 4000;
  let total = messages.reduce((sum, m) => sum + m.content.length, 0);
  
  while (total > maxTokens * 4 && messages.length > 2) {
    messages.splice(1, 1); // Remove oldest
    total = messages.reduce((sum, m) => sum + m.content.length, 0);
  }
  
  return messages;
}

Error Handling

1. Implement Retries

async function robustChat(messages: any[], maxRetries = 3) {
  let delay = 1000;
  
  for (let i = 0; i < maxRetries; i++) {
    try {
      const response = await fetch('https://regpilot.dev/api/ai/chat', {
        method: 'POST',
        headers: {
          'X-API-Key': process.env.REGPILOT_API_KEY!,
          'Content-Type': 'application/json'
        },
        body: JSON.stringify({ messages, quality: 'balanced' })
      });
      
      if (response.ok) return response;
      
      if (response.status === 429) {
        const retryAfter = parseInt(response.headers.get('retry-after') || '60');
        await new Promise(r => setTimeout(r, retryAfter * 1000));
        continue;
      }
      
      if (response.status >= 500 && i < maxRetries - 1) {
        await new Promise(r => setTimeout(r, delay));
        delay *= 2; // Exponential backoff
        continue;
      }
      
      throw new Error(`HTTP ${response.status}`);
    } catch (error) {
      if (i === maxRetries - 1) throw error;
    }
  }
}

2. Handle Specific Errors

try {
  const response = await chat(messages);
  // Process response
} catch (error) {
  if (error.status === 401) {
    // Invalid API key
    console.error('API key invalid. Please check credentials.');
  } else if (error.status === 429) {
    // Rate limited
    console.error('Rate limit exceeded. Retry later.');
  } else if (error.status >= 500) {
    // Server error
    console.error('Server error. Contact support.');
  }
}

3. Log Errors

async function chatWithLogging(messages: any[]) {
  const startTime = Date.now();
  
  try {
    const response = await chat(messages);
    
    // Log success
    logger.info({
      action: 'chat_success',
      duration: Date.now() - startTime,
      model: response.headers.get('x-regpilot-model')
    });
    
    return response;
  } catch (error) {
    // Log error
    logger.error({
      action: 'chat_error',
      duration: Date.now() - startTime,
      error: error.message
    });
    
    throw error;
  }
}

Cost Optimization

1. Monitor Spending

// Check usage regularly
const stats = await fetch(
  `https://regpilot.dev/api/usage?projectId=${projectId}`,
  { headers: { 'X-API-Key': apiKey } }
);

const data = await stats.json();

if (data.budget.remaining < 100) {
  console.warn('Low credit balance!');
  // Send alert
}

2. Set Budgets

// Enforce budget limits
const MONTHLY_BUDGET = 1000; // $1000

async function checkBudget() {
  const usage = await getUsage();
  
  if (usage.budget.used > MONTHLY_BUDGET * 0.9) {
    // 90% used - send warning
    sendAlert('Budget Warning: 90% used');
  }
  
  if (usage.budget.used >= MONTHLY_BUDGET) {
    // Budget exceeded - stop requests
    throw new Error('Monthly budget exceeded');
  }
}

3. Use Quality Tiers Wisely

function selectQuality(taskComplexity: number): string {
  if (taskComplexity < 3) return 'cheap';
  if (taskComplexity < 7) return 'balanced';
  return 'frontier';
}

Compliance

1. Enable Governor for Sensitive Use Cases

// Legal, medical, HR, or high-risk content
const response = await fetch('https://regpilot.dev/api/ai/chat', {
  method: 'POST',
  headers: {
    'X-API-Key': process.env.REGPILOT_API_KEY!,
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    messages: [...],
    governorMetadata: {
      actionType: 'legal_advice',
      recipientCountry: 'US',
      senderId: 'user_123'
    }
  })
});

2. Monitor Alerts

// Check alerts daily
async function checkAlerts() {
  const alerts = await getActiveAlerts();
  
  const critical = alerts.filter(a => a.severity === 'CRITICAL');
  
  if (critical.length > 0) {
    // Send immediate notification
    sendUrgentAlert(critical);
  }
}

3. Maintain Audit Trails

// Store audit IDs for compliance
const auditId = response.headers.get('x-governor-audit-id');

await db.conversations.update({
  where: { id: conversationId },
  data: { 
    governorAuditId: auditId,
    lastValidated: new Date()
  }
});

Development Workflow

1. Separate Environments

# .env.development
REGPILOT_API_KEY=sk_dev_xxx

# .env.staging
REGPILOT_API_KEY=sk_staging_xxx

# .env.production
REGPILOT_API_KEY=sk_prod_xxx

2. Test Before Deploy

// Test script
async function testApi() {
  const tests = [
    { name: 'Simple query', quality: 'cheap' },
    { name: 'Complex query', quality: 'frontier' },
    { name: 'With Governor', hasGovernor: true }
  ];
  
  for (const test of tests) {
    try {
      await chat([{ role: 'user', content: test.name }]);
      console.log(`✅ ${test.name} passed`);
    } catch (error) {
      console.error(`❌ ${test.name} failed:`, error);
    }
  }
}

3. Monitor in Production

// Production monitoring
setInterval(async () => {
  const health = await checkHealth();
  
  if (!health.ok) {
    sendAlert('API health check failed');
  }
}, 60000); // Every minute

Deployment

1. CI/CD Integration

# .github/workflows/deploy.yml
- name: Test RegPilot Integration
  env:
    REGPILOT_API_KEY: ${{ secrets.REGPILOT_API_KEY }}
  run: npm test

2. Environment Variables

# Vercel
vercel env add REGPILOT_API_KEY

# Netlify
netlify env:set REGPILOT_API_KEY sk_xxx

# Railway
railway variables set REGPILOT_API_KEY=sk_xxx

3. Health Checks

// Add health check endpoint
app.get('/health', async (req, res) => {
  try {
    await testRegPilotConnection();
    res.json({ status: 'healthy' });
  } catch (error) {
    res.status(500).json({ status: 'unhealthy', error: error.message });
  }
});

Monitoring

1. Track Key Metrics

const metrics = {
  requests: 0,
  errors: 0,
  cacheHits: 0,
  totalCost: 0,
  avgLatency: 0
};

// Update after each request
function recordMetrics(response: Response) {
  metrics.requests++;
  metrics.totalCost += parseFloat(response.headers.get('x-credits-charged') || '0');
  
  if (response.headers.get('x-cache-status') === 'HIT') {
    metrics.cacheHits++;
  }
}

2. Set Up Alerts

// Alert on anomalies
if (metrics.errors / metrics.requests > 0.05) {
  // Error rate > 5%
  sendAlert('High error rate detected');
}

if (metrics.cacheHits / metrics.requests < 0.1) {
  // Cache hit rate < 10%
  sendAlert('Low cache hit rate');
}

3. Dashboard Integration

// Send to monitoring service
await prometheus.gauge('regpilot_requests_total', metrics.requests);
await prometheus.gauge('regpilot_cost_usd', metrics.totalCost);
await prometheus.gauge('regpilot_cache_hit_rate', 
  metrics.cacheHits / metrics.requests
);

Related: Error Handling | Troubleshooting

Getting Started

Core Features

AI Gateway

EU AI Act Compliance

API Reference

Integration Guides

Best Practices

Advanced

​Best Practices

​Security

​1. Protect API Keys

​2. Use Secure Headers

​3. Rotate Keys Regularly

​Performance

​1. Choose Right Quality Tier

​2. Leverage Caching

​3. Optimize Prompts

​4. Manage Context Windows

​Error Handling

​1. Implement Retries

​2. Handle Specific Errors

​3. Log Errors

​Cost Optimization

​1. Monitor Spending

​2. Set Budgets

​3. Use Quality Tiers Wisely

​Compliance

​1. Enable Governor for Sensitive Use Cases

​2. Monitor Alerts

​3. Maintain Audit Trails

​Development Workflow

​1. Separate Environments

​2. Test Before Deploy

​3. Monitor in Production

​Deployment

​1. CI/CD Integration

​2. Environment Variables

​3. Health Checks

​Monitoring

​1. Track Key Metrics

​2. Set Up Alerts

​3. Dashboard Integration

Best Practices

Security

1. Protect API Keys

2. Use Secure Headers

3. Rotate Keys Regularly

Performance

1. Choose Right Quality Tier

2. Leverage Caching

3. Optimize Prompts

4. Manage Context Windows

Error Handling

1. Implement Retries

2. Handle Specific Errors

3. Log Errors

Cost Optimization

1. Monitor Spending

2. Set Budgets

3. Use Quality Tiers Wisely

Compliance

1. Enable Governor for Sensitive Use Cases

2. Monitor Alerts

3. Maintain Audit Trails

Development Workflow

1. Separate Environments

2. Test Before Deploy

3. Monitor in Production

Deployment

1. CI/CD Integration

2. Environment Variables

3. Health Checks

Monitoring

1. Track Key Metrics

2. Set Up Alerts

3. Dashboard Integration