🚀 Tokyo — AI Tracking SDK

Status: ✅ Production-Ready

Register : app.usetokyo.com/register

Tokyo SDK is a TypeScript-first client that enables seamless integration with multiple AI providers (OpenAI, Google Gemini, Anthropic Claude) using APIs that mirror their official SDKs — with automatic usage tracking and multi-tenant support.

✅ Supported Providers

| Provider | Status | Main Method | Example | |----------|--------|-------------|---------| | OpenAI | ✅ Production-Ready | client.responses.create() | client.responses.create({ model: "gpt-4o", messages: [...] }) | | Google Gemini | ✅ Production-Ready | client.models.generateContent() | client.models.generateContent({ model: "gemini-2.0-flash", contents: "..." }) | | Anthropic Claude | ✅ Production-Ready | client.messages.create() | client.messages.create({ model: "claude-3-sonnet", messages: [...] }) |

🎯 Key Features

• 🔌 Native API compatibility – identical syntax to official provider SDKs
• 🔑 Unified authentication – one Tokyo key for all providers
• 💰 Automatic billing – built-in cost tracking per customer
• ⚡ TypeScript-first – full type safety and autocompletion
• 🔄 Cross-provider fallback – retry on alternate providers automatically
• 📊 Complete usage metrics – tokens, costs, request IDs, and more

📦 Installation

npm install tokyo-ai

🔧 Quick Setup

import { Tokyo, AI_PROVIDERS } from "tokyo-ai";

const client = new Tokyo(
  AI_PROVIDERS.openai, // or gemini, anthropic
  "sk_tokyo_your_key",
  { externalCustomerId: "your-customer-id" }
);

🌟 Why Choose Tokyo SDK?

🚀 Zero Code Changes

Drop-in replacement for native SDKs. Just update your import:

// Before (Native OpenAI)
import OpenAI from "openai";
const openai = new OpenAI({ apiKey: "sk-..." });

// After (Tokyo SDK)
import { Tokyo, AI_PROVIDERS } from "tokyo-ai";
const openai = new Tokyo(AI_PROVIDERS.openai, "sk_tokyo_...", { externalCustomerId: "123" });

// ✅ Same API, no logic changes

📊 Built-In Usage Tracking

• Automatic token and cost tracking
• Real-time metrics in the Tokyo dashboard
• Per-customer billing out of the box

🔑 Unified Key Management

• One Tokyo key for OpenAI, Gemini, Claude, and more
• Keep provider API keys hidden and secure
• Fine-grained permissions and limits

💡 Usage Examples

OpenAI API

const openai = new Tokyo(AI_PROVIDERS.openai, key, config);

const res = await openai.responses.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Tell me a joke." }]
});

console.log(res.choices[0].message.content);
console.log("Cost:", res.usage.cost);

Google Gemini

const gemini = new Tokyo(AI_PROVIDERS.gemini, key, config);

const res = await gemini.models.generateContent({
  model: "gemini-2.0-flash",
  contents: "Explain AI in one sentence."
});

console.log(res.candidates[0].content.parts[0].text);

Anthropic Claude

const claude = new Tokyo(AI_PROVIDERS.anthropic, key, config);

const res = await claude.messages.create({
  model: "claude-3-sonnet",
  messages: [{ role: "user", content: "What is quantum computing?" }]
});

console.log(res.content[0].text);

OpenAI Embeddings

const res = await openai.embeddings.create({
  model: "text-embedding-3-small",
  input: "Hello, world!"
});

console.log("Embedding length:", res.data[0].embedding.length);

🔄 Provider Switching

client.updateProvider(AI_PROVIDERS.gemini); // Switch dynamically

Or create multiple clients:

const openai = new Tokyo(AI_PROVIDERS.openai, key, config);
const gemini = new Tokyo(AI_PROVIDERS.gemini, key, config);

🛠 Advanced Configuration

const config = {
  externalCustomerId: "your-customer-id", // Required
  timeout: 30_000,
  retries: 3,
  debug: true,
  headers: { "X-Custom-Header": "value" }
};

🚀 Getting Started

1. Register at usetokyo.com
2. Create a project and configure providers
3. Generate a Tokyo SDK key (sk_tokyo_...)
4. Install and import tokyo-ai
5. Use APIs with zero code changes

🚧 Limitations

• Streaming: not yet supported (in development)
• File uploads: partial support

🔮 Roadmap

• ✅ OpenAI full API (incl. embeddings & images)
• ✅ Gemini API
• ✅ Anthropic API
• ✅ Automatic billing & multi-tenancy
• 🔄 Streaming responses (coming soon)
• 🔄 Response caching
• 🔄 Advanced analytics

📞 Support

• 🌐 Website: usetokyo.com
• 📧 Email: [email protected]

📄 License

MIT License — see LICENSE.