6-Part Tutorial Series

OpenClaw on Your VPS

From a fresh Ubuntu VPS to a hardened, 24/7 AI agent connected to your messaging apps — with curated skills, scheduled automations, and integrations into GitHub, Gmail, and Nextcloud.

Autonomous AI agent

~3 hours total

1 VPS required

Start Part 1

What You'll Build

Foundation & Channels

• Hardened Gateway with Nginx SSL and systemd auto-restart
• Telegram, Discord, Slack, and WhatsApp integrations
• Flexible LLM configuration (API, OpenRouter, or local Ollama)

Skills & Automation

• Curated skills for email, GitHub, shell, and monitoring
• Cron-based schedules for briefings and health checks
• Webhook automations and multi-agent team configurations

Series Roadmap

Part 1: What It Is and What You Need

15 min

OpenClaw's architecture explained (Gateway, channels, agents, skills), why a VPS beats a laptop, hardware requirements, LLM provider options, and a roadmap of the series.

Part 2: Installation and Security Hardening

35 min

Non-root service user, UFW firewall locking down port 18789, fail2ban, Node.js 24, native and Docker install paths, systemd service, Nginx reverse proxy with SSL, gateway auth token, and credential permissions.

Part 3: Messaging Channel Integrations

25 min

Step-by-step for Telegram, Discord, Slack, and WhatsApp. BotFather setup, Message Content Intent, Bolt SDK webhooks, Baileys session persistence, and multi-channel routing to separate agents.

Part 4: LLM Configuration and Model Strategy

25 min

Direct Anthropic setup with model comparison, OpenRouter as a unified API layer with failover, and Ollama for fully local inference including RAM requirements and remote GPU server setup.

Part 5: Skills, Cron Jobs, and Automation

30 min

Safe skill installation practices (12-20% of ClawHub skills contain malicious code), curated recommendations, cron job configuration, webhook-triggered automations, and Gmail Pub/Sub for real-time email events.

Part 6: Advanced Use Cases and Integrations

35 min

VPS monitoring via Telegram with threshold alerting, email management pipeline, GitHub workflow automation, Nextcloud WebDAV integration, Tailscale zero-public-port deployment, and multi-agent team configurations.

Prerequisites

• A RamNode VPS — 4 GB RAM ($12/mo) recommended; 8 GB+ if running local LLMs via Ollama
• Ubuntu 22.04 or 24.04 LTS
• A domain name pointed at your VPS (for HTTPS access to the Control UI)
• SSH access to the server
• An LLM API key (Anthropic, OpenAI, or OpenRouter) — or plan to use Ollama locally
• At least one messaging account (Telegram recommended for first setup)

Looking for a Quick-Start Guide?

If you want a condensed, single-page walkthrough to get OpenClaw running quickly, check out our standalone deployment guide.

OpenClaw Quick-Start Guide