Node.js & TypeScript SDK
AVAILABLEThe Naagmani Node.js SDK (@naagmani/sdk) delivers typed TypeScript definitions, complete OpenAI client drop-in compatibility, and full support for both browser and Node.js runtimes.
Installation #
bash
npm install @naagmani/sdk
# or
pnpm add @naagmani/sdk
# or
yarn add @naagmani/sdk1. Basic Chat Completion #
typescript
import { Naagmani } from '@naagmani/sdk';
const naagmani = new Naagmani({
apiKey: process.env.NAAGMANI_SERVICE_TOKEN || 'nst_live_9b2d8819...',
baseURL: 'http://localhost:8080/v1', // or https://gateway.naagmani.app/v1
});
async function run() {
const completion = await naagmani.chat.completions.create({
model: 'gpt-4o',
messages: [
{ role: 'system', content: 'You are an expert TypeScript architect.' },
{ role: 'user', content: 'What are branded types in TypeScript?' }
],
temperature: 0.5,
});
console.log(completion.choices[0].message.content);
console.log('Attempt Trace ID:', completion.system_fingerprint);
}
run().catch(console.error);2. Server-Sent Events (SSE) Streaming #
typescript
const stream = await naagmani.chat.completions.create({
model: 'claude-3-5-sonnet-20241022',
messages: [{ role: 'user', content: 'Write a Fibonacci sequence function.' }],
stream: true,
});
for await (const chunk of stream) {
const content = chunk.choices[0]?.delta?.content || '';
process.stdout.write(content);
}3. Drop-in OpenAI Compatibility #
If your application already uses the official openai npm package, you only need to change the baseURL and apiKey:
typescript
import OpenAI from 'openai';
const client = new OpenAI({
apiKey: 'nst_live_9b2d8819...',
baseURL: 'http://localhost:8080/v1',
});
// All standard methods work seamlessly with Naagmani Smart Routing!
const response = await client.chat.completions.create({
model: 'gpt-4o',
messages: [{ role: 'user', content: 'Hello Naagmani!' }],
});