@aethermind/agent

Lightweight SDK for real-time AI API cost monitoring.

🚀 Installation

npm install @aethermind/agent openai
# or
pnpm add @aethermind/agent openai
# or
yarn add @aethermind/agent openai

⚡ Quick Start

import OpenAI from "openai";
import { initAethermind } from "@aethermind/agent";

// 1. Initialize Aethermind (once at app startup)
initAethermind({
  apiKey: process.env.AETHERMIND_API_KEY!, // Get from dashboard.aethermind.io
  endpoint: "https://aethermindapi-production.up.railway.app",
});

// 2. Use OpenAI normally - monitoring happens automatically!
const openai = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
});

const completion = await openai.chat.completions.create({
  model: "gpt-3.5-turbo",
  messages: [{ role: "user", content: "Hello!" }],
});

// 3. View costs in real-time at dashboard.aethermind.io
console.log(completion.choices[0].message.content);

That's it! Costs are tracked automatically.

✨ Features

• 🎯 Zero Config - Works out of the box with OpenAI SDK
• ⚡ < 5ms Overhead - Imperceptible latency impact
• 🔄 Auto-Batching - Efficient telemetry transmission every 30s
• 🛡️ Resilient - Never crashes your app, even if API is down
• 📊 Real-time Dashboard - See costs as they happen
• 🔌 Multi-Provider - OpenAI, Anthropic (more coming soon)

📖 Configuration

Basic Usage

import { initAethermind } from "@aethermind/agent";

initAethermind({
  apiKey: "your-aethermind-api-key", // Required
  endpoint: "https://api.aethermind.io", // Optional
  flushInterval: 30000, // Optional: ms between flushes (default: 30000)
  batchSize: 50, // Optional: events per batch (default: 50)
  enabled: true, // Optional: enable/disable (default: true)
});

CommonJS Usage

const { initAethermind } = require("@aethermind/agent");

initAethermind({
  apiKey: process.env.AETHERMIND_API_KEY,
  endpoint: "https://aethermindapi-production.up.railway.app",
});

Configuration Options

| Option | Type | Default | Description | | --------------- | --------- | --------------------------- | ---------------------------------- | | apiKey | string | required | Your Aethermind API key | | endpoint | string | https://api.aethermind.io | API endpoint URL | | flushInterval | number | 30000 | Milliseconds between batch flushes | | batchSize | number | 50 | Maximum events per batch | | enabled | boolean | true | Enable/disable monitoring |

📚 Examples

OpenAI Integration

import OpenAI from "openai";
import { initAethermind } from "@aethermind/agent";

// Initialize once at startup
initAethermind({
  apiKey: process.env.AETHERMIND_API_KEY!,
});

// Use OpenAI normally
const openai = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
});

// All calls are automatically tracked
const response = await openai.chat.completions.create({
  model: "gpt-4",
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "What is TypeScript?" },
  ],
});

Anthropic Integration

import Anthropic from "@anthropic-ai/sdk";
import { initAethermind } from "@aethermind/agent";

// Initialize Aethermind
initAethermind({
  apiKey: process.env.AETHERMIND_API_KEY!,
});

// Use Anthropic normally
const anthropic = new Anthropic({
  apiKey: process.env.ANTHROPIC_API_KEY,
});

// Automatically tracked
const message = await anthropic.messages.create({
  model: "claude-3-opus-20240229",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello, Claude!" }],
});

Disable Monitoring in Development

import { initAethermind } from "@aethermind/agent";

initAethermind({
  apiKey: process.env.AETHERMIND_API_KEY!,
  enabled: process.env.NODE_ENV === "production", // Only track in prod
});

🔧 Troubleshooting

Events not appearing in dashboard?

1. Wait 30 seconds (default batch interval)
2. Verify API key is correct: process.env.AETHERMIND_API_KEY
3. Check endpoint URL is correct
4. Ensure enabled: true (default)

TypeScript errors?

Make sure you have @types/node installed:

npm install -D @types/node

Want to verify initialization?

The SDK logs to console when initialized:

[Aethermind] SDK initialized successfully

If disabled:

[Aethermind] SDK initialized but disabled

🎯 How It Works

1. Instrumentation - SDK patches OpenAI/Anthropic clients
2. Event Capture - Captures model, tokens, cost, latency on each call
3. Batching - Events batched locally (default: 50 events or 30s)
4. Async Transmission - Sent to Aethermind API in background
5. Zero Impact - No blocking, no errors thrown to your app

📚 Documentation

• Full Documentation
• API Reference
• Dashboard
• Examples

🤝 Support

• 🐛 Report Issues
• 💬 Discussions
• 📧 Email: [email protected]

📄 License

MIT © Aethermind Team

Built with ❤️ for developers who care about AI costs