Skip to content

Learning path

A hands-on course on how AI coding agents work inside, built around Claude Code. Go in order: each chapter uses only what the chapters before it taught.

# Chapter Level Time Needs You'll build Quiz
Understand
00 Foundations Beginner 3 min — — Quiz
01 Runtime lifecycle Beginner 4 min 00 — Quiz
02 Settings and defaults Beginner 7 min 01 A project settings file you can see load Quiz
03 Context engineering Beginner 5 min 01 — Quiz
Build
04 Permissions Intermediate 5 min 02 A deny rule you watch work Quiz
05 Hooks Intermediate 5 min 04 A hook that blocks rm -rf, and a logger Quiz
06 Skills and plugins Intermediate 5 min 03 A one-skill plugin Quiz
07 Subagents Intermediate 6 min 03, 04 A log-scanning subagent Quiz
08 MCP fundamentals Intermediate 6 min 01 A listing of a server's tools, resources and prompts Quiz
09 Build an MCP server Hands-on 6 min 04, 08 The notes MCP server, over stdio and HTTP Quiz
Secure
10 Control layers Intermediate 4 min 04, 05 — Quiz
11 MCP security Hands-on 7 min 09, 10 A hardened notes server Quiz
Ship
12 Capstone: one plugin Hands-on 4 min 05, 06, 09 The notes-kit plugin: server, skill and hook Quiz
Production
13 Build your own harness Hands-on 7 min 04, 05, 07, 08 A tiny agent host with one read-only tool Quiz
14 Agents in CI/CD Intermediate 5 min 04, 08, 10 A read-only GitHub Actions workflow Quiz
15 Agents in production Hands-on 7 min 02, 05, 10 Settings and a hook for a read-only infra agent Quiz
16 Leaving your agent running Intermediate 5 min 04, 10 A deny rule that blocks curl, and the prompt you get without it Quiz

"Time" is the reading only. Hands-on steps take longer.

For the hands-on parts you need Claude Code, a Claude subscription or Anthropic Console account (or a supported third-party provider), and Node.js LTS (v22 or later).

Accuracy

Every chapter was checked against the pages in its Sources section on 2026-10-27. If they disagree with a chapter, trust them.