omnilink-web

OmniLink Web Client SDK — integrate OmniLink AI agents into any web application.

This is the browser/TypeScript counterpart to the Python omnilink-lib package, providing a typed client for every OmniLink REST endpoint.

Install

# From npm (when published)
npm install omnilink-web

# Or install from source
git clone https://github.com/omnilink/omnilink
cd omnilink/omnilink-web
npm install && npm run build && npm link

Get Your Omni Key

Before using the SDK you need an Omni Key (omk_...). Generate one with a single click:

Get your Omni Key

Sign in (or create a free account) and click Generate API key. Copy the key — you will need it for every API call.

Quick Start

import { OmniLinkClient } from 'omnilink-web';

const client = new OmniLinkClient({ omniKey: 'omk_...' });

// Chat
const reply = await client.chat({
  agentName: 'assistant',
  prompt: 'What is 2 + 2?',
});
console.log(reply.text);

API Reference

new OmniLinkClient(config)

| Option | Type | Default | |------------|----------|--------------------------------------| | omniKey | string | required | | baseUrl | string | https://www.omnilink-agents.com | | timeout | number | 30000 (ms) | | headers | object | {} |

Chat

// Simple chat
const reply = await client.chat({
  agentName: 'my-agent',
  prompt: 'Hello!',
  engine: 'g4-engine', // g1 = Gemini, g2 = GPT-5, g3 = Grok, g4 = Claude
});

// Streaming chat (async iterable)
for await (const chunk of client.chatStream({ agentName: 'my-agent', prompt: 'Tell me a story' })) {
  process.stdout.write(chunk.text);
  if (chunk.done) break;
}

// Multi-turn with messages array
const reply = await client.chat({
  agentName: 'tutor',
  messages: [
    { role: 'user', content: 'What is gravity?' },
    { role: 'assistant', content: 'Gravity is a force...' },
    { role: 'user', content: 'How does it work on the Moon?' },
  ],
});

Speech-to-Text

// From a Blob (e.g., from MicrophoneRecorder)
const result = await client.transcribe(audioBlob);
console.log(result.text);

// With options
const result = await client.transcribe(audioBuffer, {
  mimeType: 'audio/mp4',
  language: 'fr',
});

Text-to-Speech

// Get raw response with base64 audio
const tts = await client.synthesize('Hello world');

// Get audio as a Blob
const blob = await client.synthesizeToBlob('Hello world');

// Synthesize and play immediately
await client.synthesizeAndPlay('Hello world', {
  languageCode: 'en-US',
  speakingRate: 1.2,
});

Translation

const result = await client.translate('Bonjour le monde', {
  targetLanguage: 'English',
});
console.log(result.translation);

Memory Management

// List agents with memory status
const { agents, memoryAgents } = await client.listAgents();

// Get conversation history
const memory = await client.getMemory('my-agent');

// Set conversation history
await client.setMemory('my-agent', [
  { role: 'user', parts: [{ text: 'Hello' }] },
  { role: 'model', parts: [{ text: 'Hi there!' }] },
]);

// Clear memory
await client.clearMemory('my-agent');

Agent Profiles

const profiles = await client.listProfiles();

const profile = await client.createProfile('door-controller', {
  settings: { autoLock: true },
});

await client.updateProfile(profile.id, { name: 'smart-lock' });

await client.deleteProfile(profile.id);

Other Endpoints

// Service config
const config = await client.getConfig();

// Usage stats
const usage = await client.getUsage({ limit: 50 });

// Feedback
await client.sendFeedback({ response: 'Great answer!', command: 'hello' });

Audio Module (opt-in)

The audio module provides browser-specific helpers for microphone recording and audio playback. It is a separate entry point to keep the core SDK lean.

import { MicrophoneRecorder, AudioPlayer } from 'omnilink-web/audio';

// Record from microphone
const recorder = new MicrophoneRecorder();
await recorder.start();
// ... user speaks ...
const blob = await recorder.stop();

// Play audio
const player = new AudioPlayer();
await player.playBlob(blob);
await player.playBase64(base64Data);
player.stop();
player.destroy();

React Hooks

See demos/react/ for a complete React app with custom hooks:

import { useOmniLink } from './hooks/useOmniLink';
import { useChat } from './hooks/useChat';
import { useAudioRecorder } from './hooks/useAudioRecorder';

function App() {
  const client = useOmniLink('omk_...');
  const { messages, isLoading, send } = useChat(client, {
    agentName: 'assistant',
    engine: 'g2-engine',
  });
  const { isRecording, start, stop } = useAudioRecorder();

  // ... render chat UI ...
}

Demos

| Demo | Path | Description | |------|------|-------------| | Vanilla Chat | demos/vanilla/index.html | Single-file chat demo with zero dependencies | | Vanilla Audio | demos/vanilla/audio.html | STT/TTS demo with mic recording and playback | | Robot Control | demos/vanilla/robot.html | 2D robot simulator with AI navigation + LIDAR | | React App | demos/react/ | Full React app with hooks for chat, STT, and profiles |

Running Demos Locally

# Build the SDK first
cd omnilink-web
npm install
npm run build

# Vanilla demos — open in browser
open demos/vanilla/index.html

# React demo
cd demos/react
npm install
npm run dev

Python ↔ TypeScript Method Mapping

| Python (omnilink-lib) | TypeScript (omnilink-web) | |---|---| | chat(prompt, agent_name=...) | chat({ agentName, prompt }) | | list_profiles() | listProfiles() | | create_profile(name, settings=...) | createProfile(name, { settings }) | | update_profile(profile_id, ...) | updateProfile(profileId, { ... }) | | delete_profile(profile_id) | deleteProfile(profileId) | | list_agents() | listAgents() | | get_memory(agent_name) | getMemory(agentName) | | set_memory(agent_name, conversation) | setMemory(agentName, conversation) | | clear_memory(agent_name) | clearMemory(agentName) | | transcribe(audio, mime_type=...) | transcribe(audio, { mimeType }) | | synthesize(text, ...) | synthesize(text, { ... }) | | synthesize_to_bytes(text) | synthesizeToBlob(text) | | translate(text, target_language=...) | translate(text, { targetLanguage }) | | (no equivalent) | chatStream(options) | | (no equivalent) | synthesizeAndPlay(text) |

Build

npm run build       # Build ESM + CJS + declarations
npm run dev         # Watch mode
npm run typecheck   # Type-check without emitting

License

MIT