Kenhuang Academy
  • English

      • OpenClaw Tutorial

      Welcome to the OpenClaw Tutorial. Based on OpenClaw's official materials and the open-source openclaw/openclaw repository, this guide helps you master this open-source, self-hosted personal AI agent that runs on your own machine from scratch.

      OpenClaw (formerly Clawdbot / Moltbot) is a single-process, gateway-style autonomous agent: it runs on your computer or a $5/month VPS, takes commands through chat apps you already use — WhatsApp, Telegram, Slack — and performs real tasks like shell, browser, email, and calendar operations. Memory and skills are stored as plain Markdown / YAML files you can back up with Git and grep.

      Table of Contents

      Basics

      1. Introduction — Positioning, history, features, and comparisons
      2. Architecture & Features — Single-process gateway, five subsystems, Heartbeat autonomy
      3. Installation & Usage — One-line install, onboard, Dashboard, first chat

      Configuration

      1. Model Configuration — Anthropic/OpenAI/Google, local Ollama, openclaw.json
      2. Tool System — Shell/browser/email/calendar, tool policies and approval gates

      Channels & Capabilities

      1. Message Channels — WhatsApp/Telegram/Slack/Signal and more
      2. Skills — SKILL.md, ClawHub, auto-drafted skills
      3. Memory System — Markdown/YAML files, ~/.openclaw, Git backups

      Advanced

      1. Multi-Agent — Session isolation, per-workspace/sender instances
      2. Permissions & Security — Tool policies, CVE, ClawHub risks and hardening
      3. CLI & Automation — CLI, Dashboard, Heartbeat, Webhooks/Cron

      Practice

      1. Use Cases — Personal assistant, DevOps, automation
      2. Hands-on Practice — From install to a self-running Telegram assistant
      3. Resources — Official repo, community, and further reading

      Learning Path

      StageGoalChapters
      Day 1Install and run your first chatIntroduction → Installation → Model Config
      Days 2–3Understand tools, memory, securityTools → Memory → Security
      Week 1Connect chat platforms, automateChannels → Skills → CLI & Automation
      OngoingAutonomy and scaleMulti-Agent → Use Cases → Practice

      Prerequisites

      • Basics of the terminal and Git
      • A Linux / macOS / Windows machine (or a cheap VPS), Node.js 22+
      • At least one LLM access path: an Anthropic / OpenAI / Google API key, or local Ollama

      Related tutorial: OpenClaw shares lineage with Nous Research's Hermes Agent — Hermes can run hermes claw migrate to import config, skills, and memory from OpenClaw. See the Introduction for a comparison.

      Note: OpenClaw iterates extremely fast, so commands and config keys may change. Defer to the current docs in the official repo, and be sure to read the Security chapter.