Start here
Start Here — the 60-second router
You are one honest answer away from knowing exactly where to begin. No account, no sign-up, no placement test — just read the four questions below and…
You are one honest answer away from knowing exactly where to begin. No account, no sign-up, no placement test — just read the four questions below and follow the arrow.
The hook
Picture two developers opening this course on the same morning. One has never let an AI agent take a single step unsupervised. The other already has a cron job quietly triaging issues while they sleep. Send them both to "page one" and you fail them both: the first drowns, the second yawns. This router exists so that neither of those is you.
How the course is organized (plain English)
Think of this course as a graded curriculum with four tracks, not a flat stack of pages you grind top to bottom. Each track is a small contract: an entry check ("here's what you already know"), a body of study, hands-on labs, and an exit assessment. And here is the important part — you don't graduate a track by reading it. You graduate by building the thing it asks for.
Find your track
| Your answer | Go to | First stop |
|---|---|---|
| "I prompt an AI by hand (or not at all)" | T1 · Foundations | 01-prerequisites/environment-setup.md |
| "I know the six parts, never shipped a loop" | T2 · Practitioner | Part 2 · The Heartbeat (opens Day 2) |
| "I've shipped one loop" | T3 · Engineer | Part 5 · A Complete Loop (opens Day 2) |
| "I run fleets / set team policy" | T4 · Ultra-Pro | advanced/ (opens Day 3) |
Want the full map — every entry check and every exit assessment spelled out? learning-tracks.md
Check your tool is ready
Whatever track you land on, the price of admission is the same: one agent installed. Confirm it in ten seconds.
# Claude Code
claude --version# OpenCode
opencode --versionNothing installed yet? → 01-prerequisites/environment-setup.md
[!NOTE] Going deeper: if the phrase "the six parts of a loop" means nothing to you yet, relax — that is not a gap, it is a signal. It points straight at T1, and the whole vocabulary gets built up gently, one piece at a time, in 02-foundations/mental-models.md.
Check yourself
Q: You've used Claude Code every day for months, but every single run is you typing a prompt and watching it work. Which track?
Answer
T2 · Practitioner. Daily hand-driven use means the prerequisites and foundations will be fast review for you — but you have not yet built an autonomous loop, and that is precisely what T2 exists to teach. (If "the six parts of a loop" still reads as a foreign phrase, skim T1's foundations pages first; it costs you minutes.)
Try With AI
Open any repo you genuinely do not care about, and ask your agent:
"Read this repo and propose ONE task you could safely do repeatedly on a schedule, with a provable condition for when to stop."
Notice what you are not doing: you are not running anything yet. You are checking whether you can tell a real stopping condition from a vague one. That judgment — not the tool, not the prompt — is the skill this whole course is built to sharpen.
When it goes wrong
| Symptom | Cause | Fix |
|---|---|---|
| "Every track feels partly right" | Your skills are lopsided (deep on tools, new to loops) | Take the lowest track that still has something new; the entry checks are fast |
| "T1 feels too slow" | You skipped the entry check | Each track lists what you may skip — skip pages, never labs |
| "I picked T3 and I'm lost" | You're missing the Part 2–4 vocabulary | Drop back one track; the exit assessment will tell you when you're ready |
Glossary terms used on this page: loop, heartbeat, stopping condition — all defined in 02-foundations/glossary.md.
Sources: the graded-track structure follows the curriculum backbone of Panaversity's Loop Engineering: A Crash Course (S1). Full attribution: resources/sources.md.