Runtime Adapters Quickstart

@orlo/runtime-adapters gives you reusable adapter implementations while leaving platform behavior in your host app.

Design model

You provide:

  • credential resolution
  • org context
  • optional custom fetch

The adapter provides:

  • provider-specific request translation
  • normalized inference response shape

Example

ts
import { OpenAIAdapter } from '@orlo/runtime-adapters';

const adapter = new OpenAIAdapter({
  resolveCredential: async (orgId, provider) => {
    return {
      apiKey: process.env.OPENAI_API_KEY!,
      baseUrl: 'https://api.openai.com/v1',
    };
  },
});

const response = await adapter.infer({
  orgId: '00000000-0000-0000-0000-000000000001',
  model: 'gpt-4.1-mini',
  messages: [{ role: 'user', content: 'Summarize this incident.' }],
});

Best fit

Use this package if you want transport adapters without importing Orlo’s platform API or database logic.