Building Custom AI Skills with InstructLab Taxonomy
Create domain-specific AI capabilities using InstructLab's taxonomy systemâfrom writing skill definitions to generating synthetic training data and validating fine-tuned models.
You donât need a Raspberry Pi. You donât need a Mac Mini. You donât need any hardware at all. A $5/month VPS runs OpenClaw perfectly â and Oracle Cloudâs free tier means you can do it for $0.
OpenClawâs gateway requirements are minimal:
Thatâs well within the specs of the cheapest VPS tiers from any provider.
Oracle offers an always-free ARM instance with specs that dwarf what OpenClaw needs:
# Oracle Cloud free tier: ARM Ampere A1
# 4 OCPUs, 24GB RAM, 200GB storage
# Yes, really. For free. Forever.
# Create an instance via Oracle Cloud Console:
# Shape: VM.Standard.A1.Flex
# Image: Ubuntu 24.04 (aarch64)
# Storage: 50GB boot volume (plenty)After provisioning:
# SSH in
ssh ubuntu@<your-instance-ip>
# Install Docker
curl -fsSL https://get.docker.com | sh
sudo usermod -aG docker $USER
# Log out and back in
# Deploy OpenClaw
mkdir -p ~/openclaw && cd ~/openclaw
cat > docker-compose.yml << 'EOF'
version: '3.8'
services:
openclaw:
image: ghcr.io/openclaw/openclaw:latest
container_name: openclaw
restart: unless-stopped
ports:
- "3000:3000"
volumes:
- ./config:/home/node/.openclaw
- ./workspace:/home/node/.openclaw/workspace
environment:
- TZ=Europe/Amsterdam
EOF
docker compose up -d
# Run onboarding
docker exec -it openclaw openclaw onboardHetznerâs CX22 is the best value in cloud:
# CX22: 2 vCPU, 4GB RAM, 40GB SSD, 20TB traffic
# âŹ3.79/month (~$4.10)
# Create via Hetzner CLI or console
hcloud server create \
--name openclaw \
--type cx22 \
--image ubuntu-24.04 \
--ssh-key my-key
# Same Docker setup as above# Basic droplet: 1 vCPU, 1GB RAM, 25GB SSD
doctl compute droplet create openclaw \
--size s-1vcpu-1gb \
--image ubuntu-24-04-x64 \
--region ams3 \
--ssh-keys <key-fingerprint>A public-facing OpenClaw instance needs basic security:
# 1. Firewall â only allow SSH and OpenClaw control UI
sudo ufw default deny incoming
sudo ufw allow 22/tcp
sudo ufw allow 3000/tcp # Control UI (consider removing if not needed)
sudo ufw enable
# 2. SSH hardening
sudo sed -i 's/#PasswordAuthentication yes/PasswordAuthentication no/' /etc/ssh/sshd_config
sudo systemctl restart sshd
# 3. Auto-updates
sudo apt install -y unattended-upgrades
sudo dpkg-reconfigure -plow unattended-upgrades
# 4. Fail2ban
sudo apt install -y fail2ban
sudo systemctl enable fail2ban# Inside the container
docker exec -it openclaw openclaw configure --section copilot
# Set the model
docker exec -it openclaw sh -c 'cat >> /home/node/.openclaw/openclaw.yaml << EOF
models:
default: github-copilot/gpt-5-mini
EOF'
# Restart to apply
docker compose restart# Check status
docker exec openclaw openclaw status
# View logs
docker logs -f openclaw --tail 100
# Auto-restart on crash (already handled by restart: unless-stopped)
# Update to latest version
docker compose pull
docker compose up -d
# Backup config and workspace
tar czf openclaw-backup-$(date +%F).tar.gz config/ workspace/Choose VPS when:
Choose Pi when:
Oracle Cloud free tier: $0/month + $10 Copilot Pro = $10/month
Hetzner CX22: $4/month + $10 Copilot Pro = $14/month
DigitalOcean Basic: $6/month + $10 Copilot Pro = $16/month
Raspberry Pi 5 (amortized): $3/month + $10 Copilot Pro = $13/monthAll of these are under $20/month for a state-of-the-art AI agent thatâs always on, always reachable, and handles everything from code generation to home automation.
The Oracle Cloud free tier is genuinely hard to beat â 4 CPUs and 24GB RAM for free is absurd overkill for OpenClaw, but it means youâll never worry about resources.
AI & Cloud Advisor with 18+ years experience. Author of 8 technical books, creator of Ansible Pilot, and instructor at CopyPasteLearn Academy. Speaker at KubeCon EU & Red Hat Summit 2026.
Create domain-specific AI capabilities using InstructLab's taxonomy systemâfrom writing skill definitions to generating synthetic training data and validating fine-tuned models.
How to access the OpenClaw Control UI dashboard from an Azure VM â via SSH tunnel (secure) or public IP. Covers device pairing, dashboard authentication, and the browser-based management interface.
End-to-end guide to building a complete persistent memory system for your OpenClaw AI agent. Combine memory flush, hybrid search, file-backed notes, SQLite indexing, and session hooks into a cohesive knowledge architecture.