The course
Build a Claude-Code-class agent from scratch, one primitive per chapter, until you can explain every line of it.
free · in progress
The chapters, in the repo
Every chapter’s code, tests, and write-up are published in the repo as each chapter ships. swa grows by one primitive per chapter.
paid · later
The deep version
A deeper, paid version comes later, built around student-mode assessment. No price or date yet.
Every chapter has the same shape
- a mystery you have felt
- one experiment, one number on screen
- the law it reveals
- the diagram
- the code in swa
- the question you must now answer
The chapters
In build order. By chapter 12 the harness asks you to explain it.
| ch | Status | Primitive | The mystery |
|---|---|---|---|
| 0 | ◐ | Bare model call | What is the model actually receiving? |
| 1 | ○ | Tool loop | Same model, 30 points apart |
| 2 | ○ | File + shell tools, permission gate | Auto mode: who decided that was safe? |
| 3 | ○ | Context window | Why does my agent get dumber after an hour? |
| 4 | ○ | Prompt caching | Why did this session cost $14 and that one $2? |
| 5 | ○ | Instruction layering | Why did it forget the rule in my file? |
| 6 | ○ | Subagents | A huge band collapsing to a dot |
| 7 | ○ | Worktrees | Why did two agents clobber each other? |
| 8 | ○ | Loops | Why did it rewrite the same test three times? |
| 9 | ○ | Review gates | Why can't it check its own work? |
| 10 | ○ | Skills | On-demand instructions and what they cost |
| 11 | ○ | Cost, failure modes, evals | Harness Bench |
| 12 | ○ | Student mode | The harness makes you explain it |
Join the waitlist
Hear first when the deep version opens. It is the same list as the weekly digest, tagged so the course announcement reaches you.
Confirmation opens in a new tab. Every email has an unsubscribe link.