What is the main advantage of connecting an OpenClaw AI to Discord?

It allows you to access powerful, locally hosted AI models from anywhere via your smartphone or laptop using a familiar chat interface, without directly exposing your vulnerable home server ports to the public internet.

Why can't I just run the Discord bot on the same machine as my GPU?

While possible, separating them is often better for system stability and network architecture. For instance, your heavy GPU rig might be noisy and kept in a different room, while your Discord bot runs 24/7 on a lightweight, low-power machine like a Raspberry Pi.

What does it mean when the OpenClaw Gateway binds strictly to 127.0.0.1?

By default, the Gateway only accepts connections from 'localhost' (the exact machine it is running on) for security reasons. To connect a Bot running on a separate machine within your LAN, you must bypass or reconfigure this restriction so it accepts external IP requests.

Do I need a static public IP or port forwarding for my home network?

No. Since the Discord Bot initiates an outbound connection to Discord's servers and communicates only locally (LAN) with the OpenClaw Gateway, no router port forwarding or static public IPs are required.

How does the bot handle OpenClaw's token authentication remotely?

Remote connections to OpenClaw require specific authorization tokens to prevent unauthorized LAN access. The Bot script is designed to securely hold and pass this token during the WebSocket handshake, establishing a trusted bridge with the Gateway.

What happens to the bot if the OpenClaw AI server reboots?

Because the Bot and the AI are decoupled, they run independently. If the GPU machine goes down, the Bot remains online in Discord and can automatically reconnect via WebSocket once the Gateway is back up.

Can this setup be used with AI models other than Codex?

Yes. The Discord bot simply acts as a relay to the OpenClaw API. As long as OpenClaw supports the model (e.g., GPT-5.3, Llama 3, or other local open-source models), the bot can interact with it using the standard !ai command.

Is it safe to use this setup for executing server management commands?

Yes, especially because the integration includes a !approve flow. Sensitive server actions triggered by AI or scripts must be physically approved by an administrator in the Discord chat before execution, providing a strong human-in-the-loop security layer.

What are the system requirements for hosting the Discord Bot?

Since the heavy AI inferencing is done by the OpenClaw GPU machine, the Bot host is extremely lightweight. It only requires Python 3.11+, minimal RAM (e.g., 256MB), and a stable local network connection to the Gateway.

OpenClaw + Discord: How to Access Your Home AI from Anywhere

March 28, 2026

OpenClaw + Discord: Access your home AI from anywhere

table of contents

In a Nutshell

A Bot that allows you to access your home LAN’s OpenClaw AI agents from anywhere via Discord.

What it can do

Command	What it does
`!ai`	Ask the AI questions (Uses OpenClaw models like Codex / GPT-5.3 as-is)
`!memo`	Create memos while conversing with the AI, auto-saving to Obsidian
`!sum`	Attach and summarize files
`!tr ja Hello`	Translate
`!approve`	Server management approval flow (reef-bridge integration)
`!status`	Check the connection status to OpenClaw

Why I made this

OpenClaw has official Discord integration, but it assumes the Bot and Gateway run on the same machine.

In reality, you’d want to separate the GPU machine and the Bot machine. But when you do, you immediately hit some walls:

The Gateway only binds to 127.0.0.1 by default
Remote connections require token authentication
New devices require pairing approval (and if you mess up the timing, you get stuck)

All of these are specifications, not bugs, but they are completely undocumented.

This project provides the working code and documentation to overcome all of those barriers.

Architecture

Smartphone / PC (From anywhere)
    | Discord
    v
Discord Bot (Host A)  -- WebSocket -->  OpenClaw Gateway (Host B)
                          LAN                    |
                                                 v
                                          AI Agent (Codex etc.)

The Bot and OpenClaw run on separate machines. Linux or Windows, any combination works. Even if one goes down, the other stays alive.

Screenshots

Questions and Answers with AI

Execution Result Display after Approval

For more details

GitHub: discord-openclaw-relay — Setup guide, all commands, and troubleshooting
Practical Deployment Guide: The undocumented walls you will actually hit and their solutions

Requirements

Python 3.11+
Discord Bot Token
An OpenClaw Gateway reachable on the LAN

git clone https://github.com/superdoccimo/discord-openclaw-relay.git
cd discord-openclaw-relay
cp .env.example .env   # Edit and configure
pip install -r requirements.txt
python src/main.py

License

MIT