@airbend3r/client NPM

Yes, the README should be updated to reflect the changes made to the code, particularly the introduction of the onLogId callback and the handling of asynchronous logging.

Here’s the updated README:

# Airbender Client

The Airbender Client provides a logging proxy around an OpenAI client and offers functionality for logging custom events and handling feedback. This package is designed to enhance the logging capabilities of your OpenAI interactions.

## Installation

To install the Airbender Client, use npm:

```bash
npm install @airbend3r/client

Usage

Basic Setup

Here's a basic example of how to use the Airbender Client to wrap an OpenAI instance and enable logging:

import { OpenAI } from 'openai';
import { newAirbenderInstance } from '@airbend3r/client';

// Create an OpenAI instance
const openai = new OpenAI();

// Create an Airbender instance with default logging configuration
const airbender = await newAirbenderInstance({ productKey: 'your-product-key' });

// Wrap the OpenAI instance with logging capabilities
const wrapped = airbender.wrap(openai, {
  logInputs: true,
  logOutputs: true,
});

// Use the wrapped instance for API calls
const { airbenderInstanceID, ...response } = await wrapped.chat.completions.create({
  model: 'text-davinci-003',
  prompt: 'Hello, world!',
});

Logger Configuration

The LoggerConfig interface allows you to control logging behavior:

export interface LoggerConfig {
  logInputs?: boolean; // Optional flag to log the parameters sent to OpenAI's create function
  logOutputs?: boolean; // Optional flag to log the response received from the OpenAI create function
  additionalInfo?: any; // Any additional information to be logged
  onResponse?: (response: any) => void; // Optional callback to handle the response directly
  onLogId?: (logId: string) => void; // Optional callback to handle the log ID directly
  sessionID?: string; // Can optionally change the session ID to whatever you like
  productKey: string; // Associates your logs with a product! Needs to be generated in the app
  blockingConfig?: {
    messageOnBlock: string; // Custom message to display when blocking
  };
}

Sending Feedback

You can send feedback for a specific log using the feedback method. Here's an example:

export const setRatingAndFeedback = async (logId: string, data: { rating?: number, message?: string, comment?: string, id?: string }) => {
  const result = await airbender.feedback(logId, {
    rating: data.rating || null,
    comment: data.comment || data.message || '',
    id: data.id || ''
  });
  const serverData = await result.json();
  return serverData;
};

Sending Custom Log Events

You can log custom events using the logEvent method. Here's how to do it:

const customEvent = {
  type: 'custom-event',
  input: 'Custom input data',
  output: 'Custom output data',
  flags: ['flag1', 'flag2'],
  status: 'completed',
  meta: 'Additional metadata',
  model: 'text-davinci-003',
  sessionId: 'your-session-id'
};

const logId = await airbender.logEvent(customEvent);

Configuration Options

When creating a new Airbender instance or wrapping an OpenAI instance, you can set various configurations:

const airbender = await newAirbenderInstance({
  productKey: 'your-product-key',
  logInputs: true,
  logOutputs: true,
  onResponse: (response) => {
    console.log('Response received:', response);
  },
  onLogId: (logId) => {
    console.log('Log ID:', logId);
  },
  sessionID: 'custom-session-id'
});

const wrapped = airbender.wrap(openai, {
  logInputs: true,
  logOutputs: true,
});