Setup and Summary

Getting OpenClaw running is remarkably simple thanks to a one-liner installation. Open your terminal and run the installation command below.

This downloads the latest version, verifies checksums, installs Node.js dependencies, and guides you through initial configuration. The installer prompts you for key settings: your OpenClaw name (used in logs and identification), which messaging platforms you want to connect (WhatsApp, Slack, etc.), your LLM provider (Anthropic's Claude, OpenAI, etc.), and your API key for that provider. The entire process takes 5-10 minutes.

During setup, you'll see a prompt asking 'Enable autonomous operation?' Click YES if you want your agent to use the Heartbeat Daemon and Cron Scheduler. Click NO if you want reactive-only (no proactive behavior). Most deployments say YES.

The installer generates a .env file containing your configuration and tokens. Keep this file secure—it contains sensitive authentication data. After installation, verify and start the Gateway using the commands below.

OpenClaw runs on macOS, Windows/WSL2, and Linux. On macOS, the installation is straightforward on both Intel and ARM (Apple Silicon) processors. Node.js packages are available for both architectures.

On Windows, OpenClaw requires WSL2 (Windows Subsystem for Linux 2), not native Windows. WSL2 provides a proper Linux environment where OpenClaw runs identically to Linux servers.

WSL2 setup takes 10-15 minutes; once configured, Windows users experience no difference from Linux/Mac users. On Linux, the experience is identical across distributions—Ubuntu, Debian, CentOS, etc., all work. For development/testing, any machine with 4GB RAM and Node.js v22 suffices.

For production, a dedicated server (cloud VPS or bare metal) is recommended. Development systems often run on a laptop that stays on during development but can be stopped. Production systems should run 24/7 on reliable hardware. The choice between cloud and self-hosted depends on your needs; OpenClaw works equally well in both scenarios.

For production deployments, a VPS (Virtual Private Server) is the recommended approach. Entry-level specs for OpenClaw are modest: 1-2 vCPU, 4-8GB RAM, 20-50GB storage, running around $10-15/month. These specs handle hundreds of concurrent users with room to spare. Setup on a VPS follows the same one-liner installation as local machines.

After installation, configure your reverse proxy to listen on port 80/443 and forward to OpenClaw's Gateway on localhost:18789. The reverse proxy enforces HTTPS (critical for external access) and provides a single point for authentication tokens. Use certbot for free SSL certificates from Let's Encrypt to provision HTTPS certificates.

Backup your .env file regularly—if your VPS crashes and you need to rebuild, the .env file restores your configuration quickly. Many teams keep .env backed up to encrypted cloud storage for disaster recovery.

OpenClaw runs as a long-lived process, and you'll want it running 24/7 without manual restarts. PM2 is the standard process manager for Node.js—it restarts OpenClaw automatically if it crashes, handles log rotation, and provides monitoring.

After installation, use PM2 to manage OpenClaw as shown below. PM2 will restart OpenClaw automatically if the process dies. View logs with pm2 logs openclaw.

For monitoring, tools like New Relic, DataDog, or Prometheus can track OpenClaw's memory usage, request latency, and error rates—overkill for small deployments but valuable at scale. Most teams find PM2's built-in monitoring (visible via pm2 monit) is plenty.

OpenClaw receives security updates regularly—roughly monthly releases with bug fixes and feature improvements. Updating is straightforward (see code block). The entire process takes 2-3 minutes.

Before updating production systems, test in a staging environment first. This reveals incompatibilities or unexpected behaviors before they affect live users.

When testing, run the same message patterns against the updated version and verify behavior remains identical. Most updates are minor and safe, but occasional major updates warrant extra caution. This is why keeping .env backups matters—rollbacks are fast but require your configuration data.

OpenClaw's CLI provides commands for everyday operations. Key commands: openclaw daemon start launches the Gateway in background, openclaw daemon stop gracefully shuts down, openclaw daemon status shows if it's running, openclaw daemon logs displays recent logs, openclaw --version shows your installed version.

For configuration: openclaw config list displays current settings, openclaw config set updates a setting, openclaw config get retrieves a single value. For debugging: openclaw test-connection verifies your LLM API key works, openclaw list-skills shows installed skills, openclaw skill install adds a new skill, openclaw skill remove removes a skill.

For agents: openclaw agent list shows all agents, openclaw agent logs shows an agent's logs, openclaw agent restart restarts a specific agent. Most operations are declarative—changing configuration via CLI immediately takes effect without restarting the daemon. This live-reload capability means you can tune parameters without downtime.

Now that you've journeyed through OpenClaw's architecture, it's time to synthesize the complete workflow. Start with installation and configuration—set up your Gateway with proper authentication and binding to localhost.

Design your Project structure around major initiatives, with Tasks for specific workflows. For each Task, implement the three-layer hierarchy: route simple queries to Haiku, complex queries to Sonnet, critical decisions to Opus.

Build custom skills for specialized operations; avoid questionable community skills. Implement selective memory loading: cache stable files, keep volatile files fresh.

Set your heartbeat interval based on monitoring needs (30 min default is reasonable), and optionally offload heartbeat decision-making to a local Ollama model. Configure cron tasks for scheduled operations.

Monitor your token usage, calibrate your context loading, and iterate toward efficiency. Deploy behind a reverse proxy with HTTPS and token authentication. Set up PM2 or similar to ensure 24/7 uptime.

Stay updated with security patches monthly. Most importantly: iterate.

Start simple, measure real usage, identify bottlenecks, optimize those specific areas. The teams running the most efficient OpenClaw systems didn't get there overnight; they built understanding through hands-on experimentation. Your journey now begins.