ReachyClaw
Voice conversation · OpenClaw brain · Full body control
Reachy Mini App

Your OpenClaw agent, embodied.

Give your OpenClaw AI agent a Reachy Mini robot body. OpenClaw is the brain — it controls what the robot says, how it moves, and what it sees. OpenAI Realtime API handles voice I/O.

🖥️ Try with Simulator See features
🧠 OpenClaw brain 🎙️ OpenAI Realtime voice 💃 Full body control 🖥️ No robot required!
Reachy Mini Robot Dancing

Works with physical robot OR MuJoCo simulator!

🖥️ No Robot? No Problem!

You don't need a physical Reachy Mini robot to use ReachyClaw!

ReachyClaw works with the Reachy Mini Simulator, a MuJoCo-based physics simulation that runs on your computer. Watch your agent move and express emotions on screen while you talk. 

# Install simulator support
pip install "reachy-mini[mujoco]"

# Start the simulator (opens 3D window)
reachy-mini-daemon --sim

# In another terminal, run ReachyClaw
reachyclaw --gradio

🍎 Mac Users: Use mjpython -m reachy_mini.daemon.app.main --sim instead

📚 Simulator Setup Guide

What's inside

ReachyClaw makes OpenClaw the actual brain — every message, every movement, every decision.

🧠

OpenClaw is the brain

Every user message goes through your OpenClaw agent. No GPT-4o guessing — real responses with full tool access.

🎤

Real-time voice

OpenAI Realtime API for low-latency speech-to-text and text-to-speech. Voice I/O only — no GPT-4o brain.

🤖

Full body control

OpenClaw controls the robot body via action tags — head movement, emotions, dances, camera, face tracking.

👀

Vision

See through the robot's camera. Your agent can look around and describe what it sees.

🖥️

Simulator support

No robot? Run with MuJoCo simulator and watch your agent move in a 3D window.

⚡

Instant startup

No 30-second context fetch. GPT-4o is just a relay — the session starts immediately.

How it works

OpenClaw controls everything

  1. 🎤 Robot captures your voice
  2. 📝 OpenAI Realtime transcribes your speech
  3. 🧠 Your message goes to OpenClaw (the real brain)
  4. 🤖 OpenClaw responds with text + action tags like [EMOTION:happy]
  5. 💃 ReachyClaw executes the actions on the robot
  6. 🔊 Clean text goes to TTS — robot speaks while moving

Prerequisites

Choose your setup:

🧠 OpenClaw Gateway 🔑 OpenAI API Key 🐍 Python 3.11+

Option A: 🤖 Physical Reachy Mini robot
Option B: 🖥️ MuJoCo Simulator (free, no hardware!)

View installation guide

Quick start

Get ReachyClaw running with the simulator

# Clone ReachyClaw
git clone https://github.com/EdLuxAI/reachyclaw
cd reachyclaw

# Create virtual environment
python -m venv .venv
source .venv/bin/activate

# Install ReachyClaw + simulator
pip install -e .
pip install "reachy-mini[mujoco]"

# Configure (edit with your OpenClaw URL and OpenAI key)
cp .env.example .env
nano .env

# Terminal 1: Start simulator
reachy-mini-daemon --sim

# Terminal 2: Run ReachyClaw
reachyclaw --gradio