openclaw-oreilly-live-course

Course materials for the O'Reilly live course on building, deploying, customizing, and securing self-hosted AI agents with OpenClaw.

Repository layout

The repo is organized by asset type. Most assets show up across multiple modules — a workspace is used in the Module 1 opener, the Module 3 customization demo, and the Module 4 multi-agent demo — so an asset-type layout avoids duplication.

The default topology is one agent, one channel, one workspace. That is what Demos 1 through 6 use, and it is what the deployment/ guide sets up. The multi-agent/ folder is an explicit variant for the Demo 7 use case where you want multiple agents on one gateway.

Demo Index

The live demos run against a VPS prepared before the course. The artifacts below are what you can take home and reproduce yourself.

How to use this repo

This repo is a take-home companion to the live course. It bundles the deployment guide, the agent personas, and the demo artifacts so you can stand up your own OpenClaw gateway and reproduce what was demonstrated.

Prerequisites

• A Linux VPS (Ubuntu / Debian) or a local machine with Docker
• An LLM API key (Anthropic, OpenAI, or OpenRouter)
• Basic comfort with SSH and the command line

Default flow (single-agent)

• Deploy a gateway. Follow deployment/linux-vps-guide.md end-to-end. The base path uses an SSH tunnel; the optional Tailscale Serve section at the end is what was used on the demo VPS for the live course.