Configuration

Configure the Continum SDK for your use case

Configuration Options

The Continum SDK accepts the following configuration options:

import { Continum } from '@continum/sdk';

const continum = new Continum({
  // Required
  continumKey: string;
  
  // Provider API keys (at least one required)
  openaiKey?: string;
  anthropicKey?: string;
  geminiKey?: string;
  
  // Or use unified format
  apiKeys?: {
    openai?: string;
    anthropic?: string;
    gemini?: string;
  };
  
  // Optional
  defaultSandbox?: string;
  
  // Guardian configuration
  guardianConfig?: {
    enabled?: boolean;
    action?: 'BLOCK_ON_DETECT' | 'REDACT_AND_CONTINUE' | 'ALLOW_ALL';
    localOnly?: boolean;
    customPatterns?: Array<{
      name: string;
      pattern: RegExp;
      riskLevel: 'LOW' | 'MEDIUM' | 'HIGH';
    }>;
  };
  
  // Detonation configuration
  detonationConfig?: {
    enabled?: boolean;
  };
  
  // Strict mode
  strictMirror?: boolean;
});

Required Configuration

continumKey

Your Continum API key from the dashboard.

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY
});

Provider API Keys

At least one LLM provider API key is required:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,      // For OpenAI models
  anthropicKey: process.env.ANTHROPIC_API_KEY, // For Claude models
  geminiKey: process.env.GEMINI_API_KEY        // For Gemini models
});

Optional Configuration

defaultSandbox

The default sandbox to use for all audits (optional but recommended):

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'your-sandbox-slug'  // Use this sandbox by default
});

// Uses default sandbox
const response1 = await continum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Hello' }]
});

// Override per call
const response2 = await continum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Hello' }],
  sandbox: 'another-sandbox'  // Override default
});

If no defaultSandbox is set, you must specify sandbox in each call. Calls without a sandbox will not be audited.

guardianConfig

Configure pre-LLM PII protection:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  guardianConfig: {
    enabled: true,  // Enable Guardian (default: true)
    action: 'REDACT_AND_CONTINUE'  // Action when PII detected
  }
});

Guardian Actions:

• ALLOW_ALL: Log detection but don’t redact (monitoring mode)
• REDACT_AND_CONTINUE: Redact PII and continue with LLM call (default)
• BLOCK_ON_DETECT: Block LLM call if PII detected

detonationConfig

Configure post-LLM shadow auditing:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  detonationConfig: {
    enabled: true  // Enable shadow auditing (default: true)
  }
});

endpoint

Custom Continum API endpoint (for self-hosted deployments):

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  endpoint: 'https://your-continum-instance.com'
});

timeout

Request timeout in milliseconds:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  timeout: 30000  // 30 seconds
});

retries

Number of retry attempts for failed requests:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  retries: 3  // Retry up to 3 times
});

Environment-Specific Configuration

Development

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'dev-sandbox',
  guardianConfig: {
    enabled: false  // Disable for faster iteration
  }
});

Production

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'prod-sandbox',
  guardianConfig: {
    enabled: true,
    action: 'REDACT_AND_CONTINUE'
  },
  detonationConfig: {
    enabled: true
  }
});

Per-Call Configuration

Override configuration for individual calls:

const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'pii_strict'
});

// Override sandbox for this call
const response = await continum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Hello' }],
  sandbox: 'security_audit',  // Use different sandbox
  guardianEnabled: false,     // Disable Guardian for this call
  metadata: {
    userId: 'user_123',
    feature: 'chat'
  }
});

Multiple Instances

Create multiple Continum instances for different use cases:

// Instance for customer support (strict PII protection)
const supportContinum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'support-sandbox',
  guardianConfig: {
    enabled: true,
    action: 'REDACT_AND_CONTINUE'
  }
});

// Instance for code generation (security focus)
const codeContinum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'code-sandbox',
  guardianConfig: {
    enabled: false
  }
});

// Use appropriate instance
const supportResponse = await supportContinum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Help me with my account' }]
});

const codeResponse = await codeContinum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Write a function to sort an array' }]
});

Configuration Best Practices

Use Environment Variables

Never hardcode API keys:

// ❌ Bad
const continum = new Continum({
  continumKey: 'co_abc123...',
  openaiKey: 'sk-abc123...'
});

// ✅ Good
const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY,
  openaiKey: process.env.OPENAI_API_KEY
});

Validate Configuration

Check configuration at startup:

function createContinum() {
  if (!process.env.CONTINUM_KEY) {
    throw new Error('CONTINUM_KEY environment variable is required');
  }
  
  if (!process.env.OPENAI_API_KEY && !process.env.ANTHROPIC_API_KEY) {
    throw new Error('At least one LLM provider API key is required');
  }
  
  return new Continum({
    continumKey: process.env.CONTINUM_KEY,
    openaiKey: process.env.OPENAI_API_KEY,
    anthropicKey: process.env.ANTHROPIC_API_KEY,
    defaultSandbox: process.env.DEFAULT_SANDBOX || 'your-sandbox-slug'
  });
}

const continum = createContinum();

Singleton Pattern

Create a single instance and reuse it:

// continum.ts
import { Continum } from '@continum/sdk';

export const continum = new Continum({
  continumKey: process.env.CONTINUM_KEY!,
  openaiKey: process.env.OPENAI_API_KEY,
  defaultSandbox: 'your-sandbox-slug',
  guardianConfig: {
    enabled: true
  }
});

// app.ts
import { continum } from './continum';

const response = await continum.llm.openai.gpt_4o.chat({
  messages: [{ role: 'user', content: 'Hello' }]
});

Next Steps

OpenAI Integration

Use OpenAI models with Continum

Anthropic Integration

Use Claude models with Continum

Streaming

Stream responses from LLMs

Vision

Use vision models with images