DEV Community

Assindo
Assindo

Posted on • Originally published at assindo.com

How to Install OpenClaw (Or Skip the Setup Entirely)

OpenClaw is the most popular open-source AI agent framework, with over 145,000 GitHub stars. It can browse the web, manage files, run code, and connect to messaging platforms like WhatsApp, Telegram, and Discord. But before it does any of that, you need to install it. Here's what that actually looks like.

What You Need Before You Start

Before touching OpenClaw, you need the following ready:

  • A computer that stays on. OpenClaw runs as a background service. If your machine sleeps or shuts down, your agent goes offline. Most people use a dedicated Mac Mini ($599+), a Linux server, or a VPS ($10-40/month).
  • Docker Desktop (macOS/Windows) or Docker Engine + Docker Compose v2 (Linux). Docker is required. There is no way around this.
  • At least 4GB of RAM available for the containers, plus 10GB of disk space for images and logs.
  • An API key from at least one AI model provider (Anthropic, OpenAI, Google, or OpenRouter). This is separate from OpenClaw itself and costs money per use.
  • Terminal proficiency. You will be running shell commands, editing configuration files, and reading error logs.

If any of that sounds unfamiliar, this is going to be a long afternoon.

Step 1: Install Docker

If you don't already have Docker, install it first.

macOS: Download Docker Desktop from docker.com. Open the .dmg, drag to Applications, launch it, and wait for the Docker engine to start. You may need to grant system permissions.

Windows: Download Docker Desktop. Enable WSL 2 if prompted. Restart your computer. Launch Docker Desktop and wait for the green "running" indicator.

Linux (Ubuntu/Debian):

sudo apt-get update
sudo apt-get install docker.io docker-compose-v2
sudo usermod -aG docker $USER
Enter fullscreen mode Exit fullscreen mode

Then log out and back in for the group change to take effect.

Step 2: Clone and Build OpenClaw

Open your terminal and clone the repository:

git clone https://github.com/openclaw/openclaw.git
cd openclaw
Enter fullscreen mode Exit fullscreen mode

Now run the setup script:

./docker-setup.sh
Enter fullscreen mode Exit fullscreen mode

This script builds the gateway container image, runs the onboarding wizard, generates security tokens, and starts the service via Docker Compose. It writes configuration to ~/.openclaw/ and workspace data to ~/.openclaw/workspace.

If the automated script fails (and it sometimes does), you can build manually:

docker build -t openclaw:local -f Dockerfile .
docker compose run -rm openclaw-cli onboard
docker compose up -d openclaw-gateway
Enter fullscreen mode Exit fullscreen mode

Step 3: Configure Your AI Provider

During onboarding, you'll need to select your model provider and enter your API key. OpenClaw supports Anthropic (Claude), OpenAI (GPT), Google (Gemini), and OpenRouter.

You need to create an account with your provider, generate an API key, and paste it into the CLI wizard. Each provider charges per token, so your costs depend on how much you use the agent.

Step 4: Access the Control Panel

Once the gateway is running, open http://127.0.0.1:18789/ in your browser. You'll need to paste the gateway token (generated during setup) into Settings. If you lose the token:

docker compose run -rm openclaw-cli dashboard -no-open
Enter fullscreen mode Exit fullscreen mode

Step 5: Connect Messaging Channels

To use OpenClaw with WhatsApp, Telegram, or Discord, you need to configure each channel separately.

WhatsApp requires scanning a QR code:

docker compose run -rm openclaw-cli channels login
Enter fullscreen mode Exit fullscreen mode

Telegram requires creating a bot via BotFather and providing the token:

docker compose run -rm openclaw-cli channels add -channel telegram -token "YOUR_TOKEN"
Enter fullscreen mode Exit fullscreen mode

Discord requires creating a bot application in the Discord developer portal:

docker compose run -rm openclaw-cli channels add -channel discord -token "YOUR_TOKEN"
Enter fullscreen mode Exit fullscreen mode

Step 6: Handle Permissions and Security

The container runs as user node (uid 1000). If you hit permission errors on Linux:

sudo chown -R 1000:1000 ~/.openclaw ~/.openclaw/workspace
Enter fullscreen mode Exit fullscreen mode

For sandboxing (running agent tools in isolated containers), you need additional configuration:

scripts/sandbox-setup.sh
Enter fullscreen mode Exit fullscreen mode

This builds a separate sandbox image. You can also build variants with browser support or common dev tools, each requiring its own setup script.

Step 7: Keep It Running

OpenClaw doesn't have built-in auto-restart or monitoring. If your machine reboots, Docker restarts, or the container crashes, you need to bring it back up manually or configure Docker restart policies. There are no push notifications if something goes wrong. You check, or you don't know.

Common Problems

  • "Image missing" errors: Run the sandbox setup script, or check your Docker image tag.
  • Permission errors (EACCES): Fix uid/gid ownership on mounted directories.
  • High memory usage: OpenClaw plus Docker can consume 4-8GB of RAM depending on what the agent is doing.
  • Agent goes offline: Your computer slept, Docker Desktop closed, or the container crashed. Restart it.
  • API costs add up: You're paying per token directly to your model provider. Heavy usage can cost $20-100+/month on top of hardware costs.

Total Cost of Running OpenClaw

Let's add it up:

  • Hardware: $599+ for a Mac Mini, or $10-40/month for a VPS
  • Electricity: $5-15/month if running 24/7 at home
  • API costs: $15-100+/month depending on model and usage
  • Your time: Setup takes 1-3 hours. Maintenance is ongoing.

That's $30-155/month before you've accomplished a single task. And you're the one keeping it all running.

What OpenClaw Can't Do

Even after all that setup, OpenClaw lives entirely in the digital world. It can browse websites, write code, manage files, and chat on messaging platforms. But it cannot:

  • Make or receive real phone calls
  • Navigate IVR menus (press 1 for billing, press 2 for support)
  • Wait on hold with your insurance company
  • Send you a push notification when a task finishes

For anything that requires interacting with the real world through a phone, you're still on your own.

The Alternative: Skip All of This

Assindo is a managed AI agent that does everything OpenClaw does in the digital world, plus things it can't. It makes phone calls, navigates hold menus, posts to social media, searches the web, schedules tasks, and learns about you over time.

There's nothing to install. Download the app on iOS or Android, or use it in your browser. Your agent runs on professionally maintained cloud infrastructure with push notifications, 29 language support, and 7 AI personalities to choose from.

No Mac Mini. No Docker. No API keys. No terminal. No electricity bill. No 2 AM debugging sessions.

Starting at $20/month, everything included.


Originally published at https://assindo.com/news/how-to-install-openclaw-or-skip-the-setup

Top comments (0)