AD-RIAN Studio AI Client

Official TypeScript SDK helper for interacting with the AD-RIAN Studio AI API.

Works in Node.js, Deno, Bun, modern browsers, Next.js, Cloudflare Workers, Vercel Edge functions, and React applications (@ai-sdk/react).

Installation

Install from npm:

# npm
npm install @ad-rian-studio/ai-client

# pnpm
pnpm add @ad-rian-studio/ai-client

# yarn
yarn add @ad-rian-studio/ai-client

# bun
bun add @ad-rian-studio/ai-client

Then import it in your code:

import { createClient } from '@ad-rian-studio/ai-client';

Quick Start

1. Non-Streaming Request

import { createClient } from '@ad-rian-studio/ai-client';

const ai = createClient({
  apiKey: process.env.CHAT_API_KEY, // optional for public endpoints
});

async function main() {
  const response = await ai.sendMessage('Tell me about AD-RIAN Studio.');
  console.log('Model:', response.model);
  console.log('Answer:', response.text);
  console.log('Tokens used:', response.usage);
}

main();

2. Streaming Request

import { AdrianStudioAI } from '@ad-rian-studio/ai-client';

const ai = new AdrianStudioAI({ apiKey: process.env.CHAT_API_KEY });

async function streamResponse() {
  const response = await ai.sendMessageStream('What projects are featured?');

  // Response body contains the Vercel AI SDK UI message stream
  const reader = response.body?.getReader();
  const decoder = new TextDecoder();

  while (reader) {
    const { done, value } = await reader.read();
    if (done) break;
    console.log(decoder.decode(value));
  }
}

3. Vercel AI SDK & React Integration (@ai-sdk/react)

Use getChatTransport() to connect @ai-sdk/react's useChat directly:

import { useChat } from '@ai-sdk/react';
import { AdrianStudioAI } from '@ad-rian-studio/ai-client';

const ai = new AdrianStudioAI();

export function ChatComponent() {
  const { messages, sendMessage } = useChat({
    transport: ai.getChatTransport(),
    body: {
      variant: 'pro',
    },
  });

  return (
    <div>
      {messages.map((m) => (
        <div key={m.id}>
          <strong>{m.role}:</strong>{' '}
          {m.parts.map((p) => (p.type === 'text' ? p.text : '')).join('')}
        </div>
      ))}
    </div>
  );
}

4. Models, Providers & Mappings

The library exports the complete model registry and provider mappings used by AD-RIAN Studio:

import {
  modelRegistry,
  MODEL_PROVIDER_MAP,
  PROVIDER_MODELS_MAP,
  getModelById,
  getModelsByProvider,
  getModelsByCategory,
  type ProviderString,
  type ModelId,
  type ModelCategory,
} from '@ad-rian-studio/ai-client';

// Look up model metadata by ID
const model = getModelById('google-gemini-flash');
console.log(model?.label, model?.provider, model?.supportsTools);

// Get all models for a specific provider
const openrouterModels = getModelsByProvider('openrouter');

// Get all models in the 'pro' variant category
const proModels = getModelsByCategory('pro');

// Quick lookup maps
const providerName = MODEL_PROVIDER_MAP['opencode-zen-deepseek-v4-flash']; // 'opencode-zen'

Exported API

• Client: AdrianStudioAI, AdrianStudioAPIError, createClient(), DEFAULT_BASE_URL, normalizeBaseURL().
• API Types: ChatRequestBody, ChatTextResponse, ApiToolDefinition, ApiToolMap, ApiToolInputSchema, ApiToolError, AIError, AIResponse, AdrianStudioClientOptions, ChatOptions, UIMessage.
• Models & Providers: modelRegistry, MODEL_PROVIDER_MAP, PROVIDER_MODELS_MAP, getModelById(), getModelsByProvider(), getModelsByCategory(), ProviderString, ModelCategory, ModelRegistryEntry, ModelId, ModelString.
• Tool Utilities: isMasterTool(name), stripMasterPrefix(name), masterName(name), createToolError(msg, retryable).
• Adapters: createAdrianStudioTransport(options).

Error Handling

Non-2xx responses from the API throw an AdrianStudioAPIError:

import { AdrianStudioAPIError, createClient } from '@ad-rian-studio/ai-client';

const ai = createClient({ apiKey: process.env.CHAT_API_KEY });

try {
  await ai.sendMessage('Hello!');
} catch (error) {
  if (error instanceof AdrianStudioAPIError) {
    console.error(`Request failed (${error.status}):`, error.message);
  }
}

Development

• pnpm install — install dependencies
• pnpm typecheck — type check without emitting files
• pnpm build — build dual ESM/CommonJS output with TypeScript declarations in dist/
• pnpm test — run offline unit tests via Vitest
• pnpm test:watch — run unit tests in interactive watch mode
• pnpm test:live -- --api-key <YOUR_KEY> — run live integration tests against the live API
• pnpm test:all — run all unit and live tests

License

MIT