Assistino

A TypeScript wrapper for the OpenAI Assistants API with multi-user support and cost optimization.

Installation

npm install assistino

Usage

import Assistant from 'assistino';

const apiKey = 'YOUR_API_KEY_HERE';
const assistantId = 'YOUR_ASSISTANT_ID_HERE';

const assist = new Assistant(apiKey, assistantId, 'gpt-3.5-turbo', {
  maxMessagesPerThread: 20,
  inactivityTimeout: 60 * 60 * 1000 // 1 hour
});

async function chatExample() {
  try {
    const userId = 'user123';
    const response = await assist.chat("Hello, how are you?", userId);
    console.log("Assistant's response:", response);

    // Cleanup inactive threads (call this periodically)
   // await assist.cleanupInactiveThreads();
  } catch (error) {
    console.error("An error occurred:", error);
  }
}

chatExample();

Configuration

When creating an instance of the Assistant class, you can provide the following options:

• apiKey: Your OpenAI API key
• assistantId: The ID of your OpenAI Assistant
• model (optional): The GPT model to use (default: 'gpt-3.5-turbo')
• options (optional): An object with the following properties:
  • maxMessagesPerThread: Maximum number of messages to retain per thread
  • inactivityTimeout: Time in milliseconds after which an inactive thread is cleaned up

Methods

• chat(message, userId): Send a message and get a response
• removeUserThread(userId): Manually remove a user's conversation thread
• cleanupInactiveThreads(): Remove all inactive threads

Best Practices

• Call cleanupInactiveThreads() periodically to manage long-running applications
• Adjust maxMessagesPerThread and inactivityTimeout based on your specific use case and budget constraints

License

MIT License

Contributing

Contributions are welcome!