Deploying OpenClaw – The Ultimate One‑Click AI Assistant Guide

OpenAI, Anthropic, Google Gemini, Groq, Mistral, Ollama – all the major LLMs can now talk to you through the same, lightweight gateway. OpenClaw, an open‑source project released under MIT, bundles the gateway, a rich command‑line interface, and an optional desktop manager powered by Tauri. The OpenClawInstaller repo on GitHub gives you a ready‑to‑run script that detects your OS, installs dependencies, pulls the latest OpenClaw binaries, wires up your preferred model and chat channels, and launches the service.

Why OpenClaw? * 1‑click CLI install on macOS, Linux, and even Windows via WSL * Plug‑and‑play support for Telegram, Discord, WhatsApp, WeChat, Slack, Feishu, iMessage – no server‑side code required * Run almost any major LLM, locally with Ollama or via API, complete with custom proxy support * Persistent long‑term memory, scheduled tasks, remote command execution, and skill‑based extensions written in Markdown * Docker‑friendly so you can containerise the gateway for isolation and scalability

Below you’ll find a step‑by‑step walkthrough, from prerequisites to advanced configuration.

1. Prerequisites

Tip: If you use a managed cloud instance, consider a 2‑CPU, 4 GB RAM VM. Docker adds ~500 MB overhead.

2. One‑Click Command Line Installation

curl -fsSL https://raw.githubusercontent.com/miaoxworld/OpenClawInstaller/main/install.sh | bash

The script will: 1. Detect your OS and install required packages (Node, Docker if needed). 2. Download the pre‑built binaries for the latest OpenClaw version. 3. Execute the config‑menu.sh wizard to guide you through: * Selecting the LLM provider and model * Providing API keys or choosing a local model (Ollama) * Configuring your message channels 4. Optionally launch the gateway as a background service.

Security note: If you encounter permissions errors on macOS, try running sudo or granting the terminal full disk access.

3. Manual Clone & Setup

If you prefer to review the source or use a custom Docker Compose, follow these commands:

# Clone the repo
git clone https://github.com/miaoxworld/OpenClawInstaller.git
cd OpenClawInstaller

# Make scripts executable
chmod +x install.sh config-menu.sh

# Run the installer
./install.sh

Optional Docker deployment:

docker compose up -d

The Docker image pulls the openclaw binary and exposes port 8000 for the gateway API.

4. Quick-start Configuration

After installation, the installer usually asks if you want to start the gateway immediately. If you chose yes, you’ll see a status page: openclaw gateway status. If not, start it manually with:

# In the foreground (for debugging)
openclaw gateway

# Or as a daemon
openclaw gateway start

Configuring Messaging Channels

OpenClaw’s wizard (config‑menu.sh) exposes a menu under Message Channels:

After configuration, restart the gateway:

openclaw gateway restart

5. Advanced Settings

OpenClaw’s global configuration lives in ~/.openclaw/openclaw.json and .openclaw/env. For most users, the wizard handles this automatically. However, if you want to tweak behavior:

• Custom LLM endpoints – For Anthropic or OpenAI proxies, set ANALPHA_BASE_URL and OPENAI_BASE_URL. The script will add a custom provider to the JSON file.
• Memory & File Access – By default, enable_shell_commands and enable_file_access are false. You can enable them in the security section to allow the assistant to run shell commands or read/write files.
• Sandbox mode – Setting sandbox_mode: true confines all external calls to a safe environment.

Use the command‑line tool to adjust:

openclaw config set security.enable_shell_commands true
openclaw config set sandbox_mode true

6. Data Management & Backup

# Export chat history to JSON
openclaw export --format json > conversations.json

# Clear long‑term memory
openclaw memory clear

# Backup whole config and logs
openclaw backup

Backups are stored in ~/.openclaw/backups/. You can also version‑control the backup directory if you want an audit trail.

7. Uninstall

If you ever need to clean up:

# Stop the service
openclaw gateway stop

# Remove global npm package
npm uninstall -g openclaw

# Delete configuration files
rm -rf ~/.openclaw

8. Troubleshooting Checklist

9. Recap

OpenClawInstaller turns a bare metal machine or a Docker‑enabled VM into a full‑featured AI assistant in just a couple of minutes. With one‑click installation, a wizard‑driven configuration, and support for almost every LLM and channel, you can replace external bot platforms and retain full control over data and performance.

Next steps – Explore the OpenClaw Manager desktop app for a GUI view or dive into the OpenClaw skill system by adding custom Markdown files in ~/.openclaw/skills/.

Happy hacking, and enjoy your personal AI assistant!