Getting Started

1. Install OmniAgent Desktop

Download the installer for your platform from the download page.

Supported platforms: Windows (x64), macOS (ARM64 / x64), Linux (x64).

2. Activate License

Open OmniAgent Desktop and navigate to the License page. You'll see your machine's registration code (REG-XXXXXXXX). Send this code to your account manager to receive an activation code.

3. Load a Model

Go to the Models page. You can either:

  • Use the bundled Qwen3 model (auto-downloaded on first launch)
  • Download a model from HuggingFace by entering the repo ID and filename
  • Configure a cloud provider (OpenAI, Anthropic, etc.) in the Providers page

4. Start the Server

On the Dashboard, click "Start Server". The server will start on the configured port (default: 1234) and load the selected model.

5. Test in Playground

Go to the Playground page and send a message. The AI will respond using the loaded model.

6. Integrate into Your App

Install the frontend SDK:

npm install @omniagent/core @omniagent/react
# or
npm install @omniagent/core @omniagent/vue

Define your business tools and connect to the server:

import { defineTool } from '@omniagent/core';
import { useOmniAgent } from '@omniagent/react';
import { z } from 'zod';

const myTool = defineTool('query_data', {
  description: 'Query business data',
  parameters: z.object({
    query: z.string(),
  }),
  async execute(args) {
    const res = await fetch('/api/data?q=' + args.query);
    return { title: 'Results', output: await res.text() };
  },
});

function App() {
  const { run, events } = useOmniAgent({
    serverUrl: 'http://localhost:1234',
    tools: [myTool],
  });
  // ...
}