Configuration

Centralized configuration management with environment validation. Load settings from .env files, validate required values, and access configuration throughout your application.

What Problem Does This Solve?

Problem: Config scattered, validation missing, errors at runtime Solution: Centralized, validated configuration

Basic Usage

import { loadConfig } from 'clear-ai-v2/shared';

const config = loadConfig();

console.log(config.llm.provider);     // 'openai'
console.log(config.llm.model);        // 'gpt-3.5-turbo'
console.log(config.memory.enabled);   // true
console.log(config.tools.api_base_url);  // 'http://localhost:4000'

Configuration Structure

interface Config {
  env: 'development' | 'production' | 'test';
  llm: {
    provider: string;
    model: string;
    temperature: number;
    max_tokens: number;
    fallbackProviders?: string[];
  };
  memory: {
    enabled: boolean;
    neo4j?: { uri, user, password };
    pinecone?: { apiKey, index };
    embedding: { provider, model, dimensions };
  };
  tokens: {
    maxTokens: number;
    budgetPerOperation: number;
  };
  tools: {
    api_base_url: string;
  };
  observability?: {
    langfuse?: { publicKey, secretKey };
  };
}

Environment Variables

See Environment Setup Guide for complete list.

Validation

Config is validated on load:

try {
  const config = loadConfig();
} catch (error) {
  console.error('Invalid configuration:', error.message);
  // Required fields missing or invalid
}

Testing

yarn test loader.test.ts  # 18 tests

---

Next: Observability