OpenClaw Raspberry Pi Installation Guide

OpenClaw is an LLM-based AI agent tool designed for Raspberry Pi, supporting command execution, programming, and third-party integrations. This guide focuses on setting up OpenClaw on **Raspberry Pi 4 or 5**, providing a powerful 24/7 AI automation server.

Prerequisites

Hardware: Raspberry Pi 5 (8GB RAM recommended) or Pi 4. At least 64GB SD card and official power supply.
OS: 64-bit Raspberry Pi OS (Lite version recommended).
Dependencies: Node.js (handled by installer) and a Telegram account for remote interaction.

Step 1: System Preparation

Ensure your system is up to date and correctly configured before starting the installation.

1.1 Update System

sudo apt-get update && sudo apt-get upgrade -y

Use raspi-config to configure your timezone, locale, and network settings if you haven't already.

Step 2: Run the One-Liner Installer

The OpenClaw installation script automatically detects your OS and installs the global npm package and systemd services.

curl -fsSL https://openclaw.ai/install.sh | bash

Alternative command if the above fails:

curl -fsSL https://openclaw.bot/install.sh | bash

Step 3: Initial Configuration

After installation, the interactive Terminal UI (TUI) will launch to guide you through the setup.

QuickStart Mode: Select "QuickStart" (Space to select, Enter to confirm).
AI Model: Choose a provider (e.g., Google Gemini) and sign in via OAuth. Select your preferred model (e.g., gemini-1.5-flash).
Chat Channel: Search for @BotFather on Telegram, run /newbot to get your token, and paste it into the TUI.
Run: Select **Hatch in TUI** to start OpenClaw. Send /start in Telegram to complete the link.

Step 4: Management & Security

Once running, you can manage OpenClaw and customize its behavior.

4.1 Core Management

Reconfigure: Run openclaw configure to restart the TUI setup.
Workspace: Configuration files are stored in ~/.openclaw/workspace/. You can edit Markdown files here to define custom memories and personality.

4.2 Security Best Practices

Run on a dedicated Pi to avoid exposing your main workstation.
Use external channels (Telegram/Discord) for interaction instead of exposing local ports.
Set up a firewall to deny all inbound connections except necessary ones (SSH).

Troubleshooting (FAQ)

Q: CMake version incompatibility?
A: Upgrade to 64-bit OS. 32-bit systems often have outdated CMake versions (3.18) while 3.19+ is required.

Q: Performance is sluggish?
A: Especially on Pi 4 2GB, the UI may lag. Access via SSH tunnel: ssh -L 8080:localhost:8080 pi@your-pi-ip or upgrade to Pi 5.

Q: Telegram bot doesn't respond?
A: Verify the token from @BotFather. Ensure your Pi has internet access. Try sending /start again.

Q: Missing context or laggy responses?
A: Use local embeddings or a faster model to reduce latency. Avoid using too many channels simultaneously.