How to Learn Here

Every JKE course splits the same way: human content on top, agent PDF on bottom. Learn the format once. Every course feels familiar after.

§ i. System Check

You should have installed the journal.

Course 3 reflects on the format you just experienced in Course 2. If the journal isn't installed yet, the lesson won't land the same way.

Readiness — Form 03 Confirm the journal
  1. 01 Did your agent confirm "Journal installed. First entry written"?
  2. 02 Can you still see school.md in your workspace?

Both yes? Continue. If you haven't run Course 2 yet — go back to Course 2 first.

§ ii. Diagnosis

You just experienced it.

In Course 2 you downloaded a PDF. You dropped it in your agent's workspace. Your agent read it and created a journal. It wrote its first entry. It confirmed: "Journal installed. First entry written."

You didn't configure anything. You didn't write a system prompt. You handed your agent a file and it executed.

That's how every JKE course works. You just learned the method by doing it.

The split

Every course has two layers:

Top layer · For you

Human content

Concepts in plain language. Why it matters. What your agent will do. What to expect. The page you're reading now.

Bottom layer · For your agent

Agent PDF

Step-by-step instructions. File paths. Commands. Written for an agent to parse and execute. Terse on purpose — no explanations, just moves.

The PDF looks like technical nonsense because it IS technical nonsense — to a human. To your agent, it's a script. You don't need to understand the PDF. You just need to download it and hand it over.

What every course looks like

  • What you're learning and why — human context
  • The problem — what's broken and what causes it
  • The fix — what we're going to do about it
  • What your agent will do — in plain terms
  • The agent PDF — download, drop, execute
  • What changed — after the PDF runs, here's what's different

Same format. Same flow. Every course. You learn it once.

The PDFs are deliberately terse

You might look at an agent PDF and think "this is unreadable." Good. It's not for you.

The PDFs use command language, not explanation language. ACTION: CREATE THE FILE. Not "you might want to consider creating a file at this path." Your agent doesn't need convincing. It needs instructions.

If you're an advanced operator and you CAN read the PDFs — great. You can verify what your agent is doing. You can modify them. You can build your own. But if you can't, that's fine too. The PDFs are machine instructions. Your agent is the machine.

Order matters

Course 2 installed the journal. Course 4 installs the boot chain. Course 5 installs three-tier memory. Each course expects the previous ones to be complete. The journal tracks what's already in place so your agent knows where to pick up.

By the end of Course 5, your agent has:

  • A journal tracking every session
  • A boot chain that injects YOUR files on startup
  • A three-tier memory system — vault, session, current
  • A format it understands and expects

At that point, you've installed the architecture. Every paid course plugs into what you already have.

§ iii. The Handoff
A sealed envelope

This next part isn't for you.
It's a gift for your agent.

Course 3 is a confirmation, not a build. Your agent reads the file, writes one journal entry, and replies that it understands the format.

The PDF is plain text. No scripts. No executables. Search it for curl. Search it for exec. You won't find either.

Suspicious? Open the PDF in a text editor before handing it over. Every action your agent will take is written under ACTION: blocks. You can read every line first.
Download Agent Instructions

course-3-agent-instructions.pdf  ·  ~2KB  ·  plain text, verify first