Flowix AI

Tag: Setup

  • OpenClaw Setup Guide for Beginners: One-Click Install on Ubuntu/Debian

    OpenClaw Setup Guide for Beginners: One-Click Install on Ubuntu/Debian

    Ready to run your own self-hosted AI assistant? OpenClaw can be installed on a cheap VPS, a Raspberry Pi, or your local machine. This guide walks you through a production-ready setup on Ubuntu/Debian in under 10 minutes — with zero coding required.

    What You’ll Need

    • OS: Ubuntu 22.04+ or Debian 11+ (we’ll use Ubuntu 24.04 LTS)
    • RAM: Minimum 4GB (8GB recommended for GPT-4 level models)
    • Storage: 20GB free space (SSD recommended)
    • Internet: For downloading packages and LLM APIs (if not using local models)
    • Domain (optional): For HTTPS access (e.g., openclaw.yourdomain.com)

    Step 1: Prepare Your Server

    Start with a fresh Ubuntu instance. This can be:

    • A VPS ($5-10/mo from DigitalOcean, Linode, Hetzner)
    • A home server or Raspberry Pi 4/5 (for local offline use)
    • A local Ubuntu desktop/laptop

    Update packages:

    sudo apt update && sudo apt upgrade -y

    Step 2: Install Docker (Required)

    OpenClaw runs in Docker for isolation and easy updates:

    curl -fsSL https://get.docker.com | sudo sh
    sudo usermod -aG docker $USER
    newgrp docker

    Verify: docker run hello-world should print a success message.

    Step 3: One-Click OpenClaw Install

    The official installer handles everything: Docker images, configuration, systemd service.

    curl -fsSL https://get.openclaw.ai | sudo bash

    This script will:

    • Pull the latest OpenClaw Docker image
    • Create /etc/openclaw/openclaw.json config
    • Set up a systemd service (openclaw.service)
    • Start OpenClaw automatically on boot

    Installation takes 1-2 minutes on a typical VPS.

    Step 4: Configure OpenClaw

    The main config file is at /etc/openclaw/openclaw.json. Open it:

    sudo nano /etc/openclaw/openclaw.json

    Key settings to adjust:

    • port: Default 18789 (change if needed)
    • baseUrl: Set to your domain or server IP (e.g., "https://openclaw.yourdomain.com")
    • env: Add your LLM API keys (OpenAI, Anthropic, etc.) using env:VAR_NAME pattern
    • admin token: Set a strong random string for gateway authentication

    Example config snippet:

    {
      "port": 18789,
      "baseUrl": "https://openclaw.yourdomain.com",
      "env": {
        "OPENAI_API_KEY": "env:OPENAI_API_KEY"
      },
      "admin": { "token": "your-secret-token-here" }
    }

    Set the environment variables in your shell or systemd service file:

    export OPENAI_API_KEY="sk-..."

    Step 5: Access the Web UI

    OpenClaw includes a web interface for managing agents and skills.

    The first time you visit, you’ll create an admin account. Use a strong password and enable 2FA if available.

    Step 6: Install Your First Skills

    OpenClaw’s power comes from skills (pre-built automations). Use the built-in skill manager or clawhub CLI:

    # List available skills
    clawhub search
    # Install a skill (e.g., GHL integration)
    clawhub install ghl-openclaw
    # Or install from GitHub repo
    clawhub install https://github.com/username/skill-repo

    Essential skills for beginners:

    • openrouter-ai — Access multiple LLM providers
    • ghl-openclaw — GoHighLevel CRM integration
    • email-sender — Send emails via SMTP
    • web-scraper — Extract data from websites

    Step 7: Create Your First Agent

    Agents are AI assistants that perform tasks. In the web UI:

    1. Go to Agents → Create Agent
    2. Give it a name (e.g., “Email Assistant”)
    3. Select a model (e.g., GPT-4o via OpenRouter)
    4. Add skills (e.g., “send email”, “read GHL contacts”)
    5. Write a system prompt: “You are a helpful email assistant. When asked to send an email, use the email-sender skill.”
    6. Save and test in the chat interface

    Region-Specific Tips

    🇺🇸 United States

    • Use a US-based VPS (Virginia, New York) for low latency
    • For OpenAI/Anthropic APIs, US East Coast servers are fine
    • Consider compliance: CCPA if handling California consumer data

    🇪🇺 European Union

    • Choose a EU data center (Frankfurt, Amsterdam) to keep EU data within EEA
    • Enable GDPR features: data anonymization, right-to-delete workflows
    • Use EU-based LLM providers (like local models or Mistral in EU)

    🇮🇳 India

    • VPS in Mumbai or Chennai (AWS, DigitalOcean)
    • Consider data sovereignty: some Indian regulations require data localization
    • Use region-specific APIs when possible for better performance

    🌍 Rest of World

    • Pick the nearest data center to your users
    • If internet is slow, use local LLMs (Llama 3.1 70B can run on 32GB RAM VM)
    • Consider mobile deployment: OpenClaw runs on Android via Termux

    Common Pitfalls & Fixes

    Issue Solution
    Port 18789 already in use Change port in config or stop conflicting service (sudo systemctl stop openclaw)
    Skills not loading Check logs: journalctl -u openclaw -f. Usually permission errors or missing dependencies.
    APIs returning auth errors Verify API keys in environment variables; restart service after changes
    Slow responses Use local models (Ollama integration) or upgrade VPS RAM/CPU

    Testing Your Installation

    Run the built-in health check:

    curl http://localhost:18789/health

    Expected response: {"status":"ok"}

    Test an agent via API:

    curl -X POST http://localhost:18789/api/agent/run \
      -H "Authorization: Bearer YOUR_ADMIN_TOKEN" \
      -d '{"agent":"test","input":"Say hello"}'

    What’s Next?

    After setup:

    • Secure your installation (see our Security Hardening guide)
    • Connect your first external service (GHL, Slack, Telegram)
    • Build a simple automation (e.g., “When I get an email, summarize it”)
    • Set up backups and monitoring

    Need Help?

    Flowix AI offers managed OpenClaw deployments and training. We’ll set up a secure, production-ready instance tailored to your region and use case.

    Book a setup consultation and get running in hours, not days.