infuseai-sdk

The official Node.js client for InfuseAI. This SDK empowers you to seamlessly integrate your custom RAG apps and Knowledge Bases into any Node.js or frontend application.

📦 Installation

npm install infuseai-sdk

🚀 Usage

1. Import and Initialize

Initialize the InfuseClient with your application credentials. You can retrieve these from your App's dashboard in InfuseAI.

import { InfuseClient } from 'infuseai-sdk';

const client = new InfuseClient({
  clientId: 'YOUR_CLIENT_ID', // Your User ID
  appId: 'YOUR_APP_ID',       // The specific AI App ID
  apiKey: 'YOUR_API_KEY',     // The API Key for this App
  theme: 'minimal'            // Optional: 'minimal', 'gradient', 'bubble', etc.
});

2. Query Your App

Send prompts to your AI using the query method. The SDK manages context retrieval (RAG) and LLM inference behind the scenes.

async function askMyAssistant() {
  try {
    const response = await client.query('What does the documentation say about deployment?');
    
    console.log('Answer:', response.response);
    
    // If your app uses a Knowledge Base, you can inspect the sources used:
    if (response.sources) {
      console.log('Sources Used:', response.sources.map(s => s.name));
    }
  } catch (error) {
    console.error('Failed to query InfuseAI:', error.message);
  }
}

askMyAssistant();

3. Embed Chat Widget

You can easily embed the full chat interface into any DOM element using the mount method.

<!-- In your HTML -->
<div id="my-chat-container" style="height: 600px; width: 400px;"></div>

<script>
  client.mount("my-chat-container");
</script>

📘 API Reference

InfuseClient

Constructor

new InfuseClient(config: InfuseConfig)

InfuseConfig Properties:

| Property | Type | Required | Description | | :--------- | :------- | :------- | :---------- | | clientId | string | Yes | Your unique user identifier from the dashboard. | | appId | string | Yes | The ID of the specific app you want to interact with. | | apiKey | string | Yes | The secret API key for authentication. | | theme | string | No | Optional chat theme (default: 'minimal'). | | baseUrl | string | No | Optional override for the API endpoint (default: Production). |

client.query(text)

Sends a prompt to the InfuseAI backend.

Signature

query(text: string): Promise<QueryResponse>

Returns

A Promise resolving to a QueryResponse object:

interface QueryResponse {
  response: string;       // The AI's generated answer
  sources?: Array<{       // Optional list of sources used for RAG
    name: string;         // Name of the source file/document
    content: string;      // Snippet of the content used
  }>;
  [key: string]: any;
}

### `client.mount(elementId)`

Embeds the chat widget into the specified DOM element.

#### Signature

```typescript
mount(elementId: string): void

Parameters

| Parameter | Type | Description | | :---------- | :------- | :---------- | | elementId | string | The ID of the DOM element where the chat widget should be rendered. |

⚠️ Error Handling

The client will throw an error if:

• Missing configuration (clientId, appId, apiKey).
• Network issues prevent reaching the API.
• The API returns 4xx or 5xx status codes (e.g., Invalid API Key).

Always wrap your calls in a try/catch block to handle these gracefully.

📄 License

ISC