AI Browser Token Optimizer SDK (JavaScript/TypeScript)

Reduce your OpenAI and Anthropic costs by 60-90% automatically.

Drop-in replacement for OpenAI and Anthropic clients that automatically compresses long text before sending it to LLMs.

Features

• ✅ Drop-in replacement - Change just 2 lines of code
• ✅ TypeScript support - Full type definitions included
• ✅ Automatic compression - No manual work required
• ✅ 60-90% token savings - Typical reduction
• ✅ Supports OpenAI & Anthropic - Works with all models
• ✅ Smart detection - Auto-detects content type
• ✅ Zero config - Works out of the box

Installation

npm install @aibrowser/optimizer

With OpenAI:

npm install @aibrowser/optimizer openai

With Anthropic:

npm install @aibrowser/optimizer @anthropic-ai/sdk

Quick Start

OpenAI (TypeScript)

import { OptimizedOpenAI } from '@aibrowser/optimizer';

// Just change these 2 lines:
const client = new OptimizedOpenAI({
  openaiKey: 'sk-...',           // Your OpenAI API key
  optimizerKey: 'your-api-key'   // Your AI Browser API key
});

// Use exactly like OpenAI:
const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [
    { role: 'user', content: 'Explain this code: ' + longCodeFile }
  ]
});

console.log(response.choices[0].message.content);
console.log(`Tokens saved: ${client.getTotalTokensSaved()}`);

OpenAI (JavaScript)

const { OptimizedOpenAI } = require('@aibrowser/optimizer');

const client = new OptimizedOpenAI({
  openaiKey: 'sk-...',
  optimizerKey: 'your-api-key'
});

const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: 'long text...' }]
});

Anthropic (TypeScript)

import { OptimizedAnthropic } from '@aibrowser/optimizer';

const client = new OptimizedAnthropic({
  anthropicKey: 'sk-ant-...',      // Your Anthropic API key
  optimizerKey: 'your-api-key'     // Your AI Browser API key
});

const response = await client.messages.create({
  model: 'claude-3-5-sonnet-20241022',
  max_tokens: 1024,
  messages: [
    { role: 'user', content: 'Analyze these logs: ' + longLogs }
  ]
});

console.log(response.content[0].text);
console.log(`Tokens saved: ${client.getTotalTokensSaved()}`);

Configuration

Compression Threshold

const client = new OptimizedOpenAI({
  openaiKey: '...',
  optimizerKey: '...',
  threshold: 5000  // Only compress if > 5000 characters
});

Disable Auto-Compression

const client = new OptimizedOpenAI({
  openaiKey: '...',
  optimizerKey: '...',
  autoCompress: false  // Disable automatic compression
});

Custom Optimizer URL

const client = new OptimizedOpenAI({
  openaiKey: '...',
  optimizerKey: '...',
  optimizerUrl: 'http://localhost:3002/v1'  // Use local instance
});

Examples

Code Explanation

import { OptimizedOpenAI } from '@aibrowser/optimizer';
import fs from 'fs';

const client = new OptimizedOpenAI({
  openaiKey: process.env.OPENAI_KEY,
  optimizerKey: process.env.OPTIMIZER_KEY
});

// Read large code file
const code = fs.readFileSync('large_codebase.js', 'utf-8');  // 10,000 lines

// Without optimization: ~30,000 tokens = $0.90
// With optimization: ~3,000 tokens = $0.09
const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: `Explain this code:\n\n${code}` }]
});

console.log(response.choices[0].message.content);
// Savings: $0.81 (90%)

Log Analysis

import { OptimizedOpenAI } from '@aibrowser/optimizer';
import fs from 'fs';

const client = new OptimizedOpenAI({
  openaiKey: process.env.OPENAI_KEY,
  optimizerKey: process.env.OPTIMIZER_KEY
});

const logs = fs.readFileSync('error.log', 'utf-8');  // 50,000 lines

const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: `Find the root cause:\n\n${logs}` }]
});

console.log(response.choices[0].message.content);

Migration from Existing Code

Before:

import OpenAI from 'openai';

const client = new OpenAI({ apiKey: 'sk-...' });

const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: '...' }]
});

After:

import { OptimizedOpenAI } from '@aibrowser/optimizer';

const client = new OptimizedOpenAI({
  openaiKey: 'sk-...',
  optimizerKey: 'your-api-key'
});

const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: '...' }]
});

Only 2 lines changed! ✅

API

OptimizedOpenAI

class OptimizedOpenAI {
  constructor(options: {
    openaiKey: string;
    optimizerKey: string;
    optimizerUrl?: string;  // Default: 'https://api.aibrowser.dev/v1'
    autoCompress?: boolean; // Default: true
    threshold?: number;     // Default: 2000
    baseURL?: string;       // OpenAI base URL
    organization?: string;  // OpenAI organization
  });

  // Get total tokens saved
  getTotalTokensSaved(): number;

  // Access to OpenAI methods (with auto-compression)
  chat: { completions: { create(...) } };
  completions: OpenAI.Completions;
  embeddings: OpenAI.Embeddings;
  images: OpenAI.Images;
  // ... all other OpenAI methods

  // Direct access to underlying OpenAI client
  raw: OpenAI;
}

OptimizedAnthropic

class OptimizedAnthropic {
  constructor(options: {
    anthropicKey: string;
    optimizerKey: string;
    optimizerUrl?: string;  // Default: 'https://api.aibrowser.dev/v1'
    autoCompress?: boolean; // Default: true
    threshold?: number;     // Default: 2000
  });

  // Get total tokens saved
  getTotalTokensSaved(): number;

  // Access to Anthropic methods (with auto-compression)
  messages: { create(...) };

  // Direct access to underlying Anthropic client
  raw: Anthropic;
}

Use Cases

Perfect for:

• 📊 Code analysis - Explain large codebases
• 🐛 Debugging - Analyze error logs
• 📚 Documentation - Process technical docs
• 🤖 AI Agents - Optimize context for agents
• 💬 Long conversations - Compress chat history
• 🔍 Research - Summarize research papers

Get API Key

Sign up at: https://aibrowser.dev/signup

Pricing

• Free tier: 10,000 compressions/month
• Pro: $9/month - Unlimited compressions
• Enterprise: Custom pricing

ROI: Average user saves $50-200/month in LLM costs, paying only $9 for the optimizer.

FAQ

Q: Does it work with ESM and CommonJS? A: Yes, both are supported.

Q: TypeScript support? A: Full TypeScript definitions included.

Q: Does it modify the LLM responses? A: No, only input text is compressed.

Q: What if compression fails? A: Falls back to original text automatically.

License

MIT License

Support

• 📧 Email: [email protected]
• 💬 Discord: Join our community
• 📖 Docs: https://docs.aibrowser.dev
• 🐛 Issues: https://github.com/yourusername/aibrowser-optimizer/issues

Made with ❤️ by the AI Browser team