Call GetAPI.ONE from Node.js

Prepare a server-side project

• A maintained Node.js release and an ESM project.
• An enabled, bounded GetAPI.ONE key.
• An exact current catalog model that declares the Responses endpoint.

npm install openai

Create the client and request

1. Inject GETAPI_ONE_API_KEY through the server process environment.
2. Set baseURL to https://www.getapi.one/v1.
3. Replace the placeholder with an exact current catalog ID.
4. Run one small request only when you accept possible billing.
5. The first check disables automatic retries. Add a bounded retry policy only after classifying safe-to-repeat operations.

import OpenAI from 'openai';

const apiKey = process.env.GETAPI_ONE_API_KEY;
if (!apiKey) throw new Error('Set GETAPI_ONE_API_KEY before starting');

const client = new OpenAI({
  apiKey,
  baseURL: 'https://www.getapi.one/v1',
  timeout: 30_000,
  maxRetries: 0,
});

const response = await client.responses.create({
  model: '<MODEL_ID_FROM_CURRENT_CATALOG>',
  input: 'Reply with one short readiness check.',
});

console.log(response.output_text);

Keep runtime boundaries clear

Verify the expected result

• The promise resolves and response.output_text contains the short reply.
• No credential appears in client-side bundles, logs, or error pages.
• A matching request or usage record is visible in the ONE console.

Fix common errors

Protect production usage

• Store secrets in the deployment platform’s secret manager, not in package.json or frontend environment variables.
• Apply request timeouts, error handling, and your own user authorization before exposing a server endpoint.
• Rotate leaked keys and review usage records for unexpected calls.

Next steps

• Add server-side streaming
• Review the security checklist