Learn Agent Engineering

A 12-course curriculum teaching AI agent development from first principles to production systems. Covers Claude Code fundamentals, tool calling, workflows, context management, multi-agent orchestration, observability, and harness design.

Live at learn.agentmentor.dev · Available in English, 中文, 日本語, 한국어, Español, Português (BR)

What You'll Learn

1. Claude Code Skills — Your first agentic prompt and how Claude Code executes it
2. Agent Workflows — Sequential, parallel, and conditional execution patterns
3. Prompt Engineering — System prompts, few-shot examples, and chain-of-thought
4. Tool Calling — Design interfaces, handle errors, and compose multi-tool chains
5. Memory & State — Persistent storage, context windows, and state machines
6. Multi-Agent Collaboration — Delegation, specialization, and coordination
7. Agent Harness — Runtime controls, safety gates, and turn budgets
8. Context Engineering — Efficient prompting under token constraints
9. State Persistence — Making long tasks survive interruptions
10. Verification — Testing agents, eval sets, and LLM-as-judge
11. Observability — Logging, tracing, and debugging agent behavior
12. Orchestration — Building graphs, review loops, and deterministic pipelines

Each course includes:

• 6 lessons with hands-on exercises
• Real code you can run locally
• Capstone projects demonstrating production patterns
• Sources cited throughout

How It Was Built

This curriculum was generated by the Agent Mentor skill — a $4.99 Claude Code skill that writes pedagogically sound courses with exercises, citations, and multi-locale glossaries. The skill was invoked 12 times (once per course) via Claude Code, producing the entire path without direct human authoring.

Why mention this? It demonstrates what systematic agent workflows can produce when properly constrained: structured learning materials that maintain consistency across 72 lessons and 6 languages. The courses teach the same techniques that built them.

The skill itself is closed-source and available at agentmentor.dev. This repository contains only its output — the learning materials — released under MIT so the community can use, translate, and improve them.

Running Locally

git clone https://github.com/fancydirty/learn-agent-engineering.git
cd learn-agent-engineering
npm install
npm run dev

Visit http://localhost:3000 and select a course. Progress is saved locally in .next/cache.

Contributing

We welcome:

• Corrections to lesson content or code examples
• Translations into additional languages
• Proposals for courses 13+ extending the path

See CONTRIBUTING.md for guidelines. For major changes, open an issue first to discuss scope.

Deployment

The canonical site runs on Cloudflare Pages via OpenNext. To deploy your own:

npm run deploy

Requires Cloudflare account and wrangler configured. See Cloudflare Pages docs for setup.

Tech Stack

• Next.js 15 — React framework with App Router
• OpenNext for Cloudflare — Deploys Next.js to Cloudflare's edge
• Markdown — Lesson content with custom fences for exercises
• Playwright — E2E tests for reader and interactive components

License

MIT License — see LICENSE for details.

---

Acknowledgments: Built with Claude Opus 5, Claude Sonnet 5, and Claude Fable 5. Course quality reviewed by adversarial agent fleets to catch translation drift and structural defects before publication.