OpenClaw Academy
Deep-dive into OpenClaw's architecture, codebase, and internals.
10 modules · 33 lessons · A community educational resource for OpenClaw
OpenClaw Overview
What OpenClaw is, how it works end-to-end, and how to think about the system as a whole.
Gateway Architecture
Deep dive into the control plane: WebSocket protocol, session management, and how the Gateway daemon works.
The Channel System
How OpenClaw connects to messaging platforms — architecture, Telegram, WhatsApp, Discord, Slack, and the broader...
Agent System
The AI runtime: LLM calls, tool execution, memory management, compaction, and multi-agent routing.
Skills & Hooks
The extensibility system: how skills are structured, loaded, and how the hook/cron system works.
Security Model
Trust boundaries, prompt injection defenses, elevated mode, and sandboxing — keeping your Gateway safe.
Configuration Deep Dive
Every major block of openclaw.json: model config, agent config, auth profiles, and the doctor tool.
Extending OpenClaw
Writing skills, custom routing, and integrating with the Model Context Protocol (MCP).
Deployment Patterns
How to run OpenClaw reliably — macOS LaunchAgent, Docker Compose, Linux VPS with systemd, HTTPS reverse proxy, and...
Case Study: Our Setup
How we built on top of OpenClaw: our Mac Mini setup, custom skills, QMD memory backend, and lessons learned.