The curriculum
Seven chapters, in order.
Each chapter is four short lessons, each with a workshop of tasks you check yourself. Start at the top; the harder chapters build on the first.
Start here · free
Chapter 1 is free. Begin with it, then decide.
4 lessons · 23 workshop tasks
Introductory
- 1
4 lessons · 23 workshop tasks
You may know Claude from the website or the app; this chapter moves you to Claude Code in the terminal, gently: what a terminal is, why every lesson here uses it, and how to undo any mistake.
Open chapter →- 1Setting up Claude Code
- 2How Claude remembers
- 3Prompting
- 4Checking Claude's work
Intermediate
- 2
- 1A check that can fail
- 2Write the evals first
- 3A separate reviewer
- 4Our own test behind a claim
- 3
- 1What a hook is
- 2The moments hooks run
- 3A hook that blocks
- 4A hook that reminds
- 4
- 1A one-page spec
- 2The first working slice
- 3A check for each step
- 4Shipping without a team
- 5
- 1What a skill is
- 2Skills that call skills
- 3What a subagent is
- 4When to hand work to one
Advanced Intermediate
- 6
- 1What an agent can reach
- 2Deny rules for secrets
- 3Running untrusted work apart
- 4Reading AI output as untrusted
- 7
- 1Split a job
- 2Orchestrator and workers
- 3What each worker is given
- 4Checking results