This repository provides an official interactive tutorial for OpenHarness, an AI Agent Harness framework powering tools like Claude Code. Aimed at developers, researchers, and power users, it demystifies OpenHarness's complex architecture, enabling rapid comprehension of core concepts like Agent Loops, tools, memory, and multi-agent systems without requiring prior AI or programming expertise. The benefit is a structured, engaging path from zero to hero in understanding next-generation AI coding tools.

How It Works

This project is a 12-chapter interactive tutorial designed as a companion to the HKUDS/OpenHarness framework. It breaks down complex AI Agent concepts using real-world analogies, animated diagrams, hands-on simulators, and embedded quizzes with instant feedback. The curriculum systematically covers fundamentals, core mechanisms (Agent Loop, Tool System, Permission System), intelligence layers (Skills, Memory, MCP Integration), and advanced topics like Multi-Agent Collaboration, offering a practical, self-paced learning experience.

Quick Start & Requirements

• Primary install/run command:

  git clone https://github.com/joyehuang/Learn-Open-Harness.git
  cd Learn-Open-Harness
  npm install
  npm run dev
  

  Then open http://localhost:3000.
• Prerequisites: Node.js (implied by npm).
• Links:
  • Live Demo: https://learn-openharness.vercel.app
  • OpenHarness Repo: https://github.com/HKUDS/OpenHarness

Highlighted Details

• Comprehensive 12-chapter curriculum covering AI Agents, Harness architecture, Agent Loop, Tool System (43+ built-in tools), Permission System (three-tier), Skills, Memory (4 types), MCP Integration, and Multi-Agent Collaboration.
• Interactive elements include an Agent Loop animation, Tool Explorer, Permission Sandbox, and Architecture Diagram.
• Features embedded quizzes with instant feedback, auto-tracked progress, dark mode, and responsive design for mobile, tablet, and desktop.
• Content is structured in easily modifiable TypeScript files, facilitating community contributions.

Maintenance & Community

The project is primarily authored by @joyehuang. Contributions are welcomed via pull requests. A link to the project's issues is provided for bug reports and feature requests. The author also notes significant traction on Xiaohongshu (32k likes).

Licensing & Compatibility

• License Type: MIT.
• Compatibility: The MIT license is highly permissive, allowing for commercial use and integration within closed-source projects without significant restrictions.

Limitations & Caveats

This repository serves as an educational resource and interactive tutorial for OpenHarness, not as a runtime framework itself. As it was developed shortly after OpenHarness's launch, users should verify its content against the latest OpenHarness versions for complete accuracy regarding evolving features.