Getting Started
Install Uplink and have your local AI chat running in under 5 minutes.
What is Uplink?
Uplink is a chat interface built specifically for OpenClaw. It runs as a Node.js web server on your machine and connects to the OpenClaw gateway — giving you a polished, feature-rich frontend for your AI assistant.
Think of it this way: OpenClaw is the brain, Uplink is the face. OpenClaw handles the AI sessions, model routing, memory, and tool execution. Uplink gives you a beautiful interface to interact with all of that — text chat, voice chat, themes, encryption, multi-device access — all self-hosted on your hardware.
- Gateway API — OpenClaw's OpenAI-compatible API powers all chat and streaming
- Session management — Conversations persist through OpenClaw's session store
- Multi-model routing — Switch between AI providers (OpenAI, Anthropic, local models) via OpenClaw's satellite system
- WebSocket streaming — Real-time token streaming through OpenClaw's gateway proxy
System Requirements
| Component | Minimum | Recommended |
|---|---|---|
| Node.js | v18+ | v20 LTS or later |
| OS | Windows 10 / macOS 12 / Ubuntu 20.04 | Latest stable release |
| RAM | 4 GB | 8 GB+ |
| GPU | Not required | NVIDIA GPU (for local TTS with Coqui XTTS). Piper TTS runs on CPU. |
| Network | LAN access | Tailscale for remote access |
Installation
Quick Start with npx (Recommended)
The fastest way to run Uplink is with npx — no clone or install step needed:
npx @mooncompany/uplink-chat
That's it — Uplink downloads and starts automatically. No build tools, no Docker, no complex setup. Open http://localhost:3456 in your browser and you're ready to go.
If you want the uplink-chat command available everywhere without npx:
npm install -g @mooncompany/uplink-chat
Then just run uplink-chat from anywhere.
Alternative: Install from Source
If you prefer to clone the repository and run from source:
git clone https://github.com/mooncompany/uplink.git
cd uplink
npm install
npm start
All dependencies are pulled in by npm. No native compilation required for the core app.
Uplink can use Sharp to automatically compress uploaded images before sending them to the AI. Sharp is optional — if it's not installed, images are sent at their original size. To enable compression:
npm install sharp
If Sharp fails to install (common on some systems without build tools), don't worry — everything else works fine without it.
You'll see output like:
✓ Uplink running at http://localhost:3456
✓ WebSocket server ready
✓ OpenClaw gateway connected
Open http://localhost:3456 in your browser and you're in.
First Run Setup
On first launch, Uplink will prompt you to configure a few things in this order:
- Your Name — Enter your display name for the chat.
- Assistant Name — Choose a name for your AI assistant.
- Gateway URL — Enter your OpenClaw gateway URL. If OpenClaw is running on the same machine, Uplink will detect it automatically. You can change this later in Settings → Connection.
- Gateway Token — Enter the gateway token found in your OpenClaw
config.yaml. - Voice toggle — Enable or disable voice responses.
- Encryption toggle — Optional. If enabled, all conversations are encrypted with AES-256. You can skip this and enable it later in Settings → Privacy.
Theme selection is available after setup — go to Settings to choose from 5 built-in themes (Midnight, Daylight, Ember, Forest, Phantom). Default is Midnight (dark with cyan accents).
Quick Verify
To verify everything is working, open the app in your browser, type a message, and confirm you get a response from the AI. If you see an error about the gateway connection, check that OpenClaw is running and accessible.
Next Steps
- Configuration CLI Reference — Customize the Settings panel and environment
- Voice Chat — Enable speech-to-text and text-to-speech
- Remote Access — Access Uplink from other devices
- Themes — Browse the 5 built-in themes (Midnight, Daylight, Ember, Forest, Phantom)