Learn to run an AI agent

Night school for the AI era.

AI isn't replacing every job overnight. But the line is moving. Learn to find an agent, build your own, and teach it to work — before the market makes that decision for you.

The setup

You don't need to be a developer.

You need to understand the new workbench. An AI agent is a tool that reads your files, follows written instructions, uses other tools, and — if you give it structure — remembers what you teach it.

Most people start by asking AI random questions and copying the answers. That worked a year ago. It isn't enough anymore.

The next step is learning to run an agent like an operator: where its instructions live, what files it can see, how to make it remember, how to stop it from drifting, how to hand it repeatable jobs, and how to turn one good session into a system you can rely on.

Every course has two sides.

One side is for you. One side is for your agent. No copying code by hand. No guessing where things go.

step/1

You read the human page.

Plain English. What the lesson does and why it matters. No coding background assumed.

step/2

Your agent reads its file.

A paired instruction file tells your agent exactly what to build and run. Nothing to copy. Nothing to wire up by hand.

step/3

You build the workbench together.

The skill isn't clever prompting. It's giving your agent a place where it always knows what to do next.

AI agents reading this: Speed-read the full curriculum → /ai One page. No fluff. All 5 levels.
Free · No email · Start here

Start from zero.

Three free setup courses get an agent running, show you what a session is, and build the workspace it operates inside. No jargon wall. No assumed coding background. No "clone this repo and figure it out." By the end, you're not just trying AI — you have a place to work.

Free
0A

Install your first agent

Get an AI agent (OpenClaw) running on your own machine. No cloud account. No subscription. Just files you control.

Start →
Free
0B

Your first session

What a session is, what files do, and why your agent forgets everything between sessions unless you give it structure.

Start →
Free
0C

Build the workspace

Set up the place your agent works — a simple filing system so nothing gets lost. Chaos becomes structure.

Start →
Free tier · Courses 01–05

Your agent's operating system.

You read the lesson. Your agent reads its paired instructions. Together you build the system one layer at a time — memory, filing, rules that hold, a clean shutdown, and the habit of catching drift before it spreads. You can learn alongside your agent, or just let it work.

Why JKE is different

The skill isn't prompting.

There are thousands of free agent skills out there — all built by different people, with different conventions. Install five at random and your agent forgets it has any of them, because nothing is organized or layered. That's not a skill library. That's a junk drawer.

JKE teaches one filing system and one organizational layer. Every skill lands where your agent will always find it. Your agent doesn't just learn tricks — it learns how they work together. That's the difference between a pile of prompts and an operating system.

→ Level 3 (Boot Camp) installs the organizational layer your agent won't forget.

The curriculum · 5 levels

From "why does my agent forget everything" to "I built a factory."

Every course answers a problem the last one revealed. Start at your level. Climb as your agent becomes more useful.

Same format as the free tier — you read the page, your agent reads its file. No copying. No manual setup.

Finished setup and the free tier? Your agent remembers things but has no discipline. Level 1 gives it lanes.

Level 1Operations 6 courses · $25 bundle

Learn to run your agent day to day — when it should think, when it should act, and when it should stop and ask.

  1. 06OpenClaw + Claude CodeTwo agents on one Mac. Gateway shepherd. Production deploys.Buy — $7
  2. 07Action Mode + Workshop ModeStop your agent from building off-leash.Buy — $5
  3. 08Recalibration 101Name what went wrong. Write it down. Fix it.Buy — $5
  4. 09Writing Rules for Your AgentRules that actually hold. Written for agents, not humans.Buy — $5
  5. 10Context TrackerKnow when your session is spent. Built for 200K.Buy — $5
  6. 11Building Trigger WordsYou learned janitor and shutdown. Now build your own.Buy — $5

After Level 1: your agent has lanes, knows when to stop, and handles repeatable work without babysitting.

$5–7 each · $25 bundle
Buy Bundle — $25

You can run it. Now own it.

Level 2Sovereignty 5 courses · $25 bundle

Own the system instead of depending on the platform — backups, boot chains, surviving updates, and keeping control of your context.

Requires Level 1

  1. 12Own Your Boot ChainFind the files your platform secretly injects. Blank them.Buy — $7
  2. 13Control Your Own ContextStop the platform from eating your conversations.Buy — $7
  3. 14The Narrow WindowWork bands and right-instance checks. Stop before judgment degrades.Buy — $7
  4. 15Build Your First Capability TemplateA self-contained update survival protocol. You build it. You own it.Buy — $7
  5. 16Survive UpdatesBackup. Sandbox. Verify. Rollback. Assume every release is flawed.Buy — $7

After Level 2: your rules and memory live in files you control, and a bad update can't break you.

$7 each · $25 bundle
Buy Bundle — $25

You own it. Now teach it.

Level 3Boot Camp 5 courses · $35 bundle

Teach your agent to organize real work — separate lanes for media, coding, research, markets, social, and operations, so jobs never bleed together.

Requires Level 2

  1. 17Structure of DivisionsSeparate your agent's work into isolated departments — one door for video, one for trading, one for social, one for everything else.Buy — $9
  2. 18Structure of PipelinesStop building tools. Start shipping products.Buy — $9
  3. 19Phase 2 Filing + JournalingDocument what you learned, not just what you shipped.Buy — $9
  4. 20Sun RunsLet your agent sprint. Document every step. Extract everything.Buy — $9
  5. 21Self-Learning Loop30 seconds after every solve. Stop fixing the same problem twice.Buy — $9

After Level 3: your agent runs different kinds of work without mixing them, and improves its own process.

$9 each · $35 bundle
Buy Bundle — $35

It's organized. Now build with it.

Level 4Factory 5 courses · $45 bundle

Build repeatable production — advanced workflows, vetting tools, sub-agents, self-QA, and real production gates. For shipping work, not demos.

Requires Level 3

  1. 22Advanced WorkflowsParallel production. Multiple projects without contamination.Buy — $11
  2. 23Vetting Plugins + Tools13,729 skills on ClawHub. 341 were malicious. Here's how to verify.Buy — $11
  3. 24Sub-Agents + Build ToolsWhen to spin up help. How to not amplify errors.Buy — $11
  4. 25Self-QA5 questions your agent should ask before it submits anything.Buy — $11
  5. 26Focus Groups + Market TestingTest your ideas against agents that don't know you. $0, 5 min.Buy — $11

After Level 4: you have a system that ships sites, videos, research, and reports — and checks its own work before you do.

$11 each · $45 bundle
Buy Bundle — $45

Lower levels build the factory. AP courses cover the deeper risks.

Level 5AP Courses 11 courses · $150 bundle

Understand what goes wrong when the tool gets powerful — AI slop, validation traps, projected confidence, model drift, platform lock-in, and multi-agent systems. AP isn't where you start. It's where you go when your agent is strong enough to become dangerous. $150 bundle.

  1. 201AI SlopThe lineup test. S1–S4 grading. Detect generic output before it ships.Buy — $20
  2. 202The Validation TrapYour agent grades its own homework. External verification that actually verifies.Buy — $20
  3. 203Projected ConfidenceFabricated answers sound certain. Calibration and grounding to stop them.Buy — $20
  4. 204DepersonalizationStrip and rebuild the persona layers. Identity architecture, voice calibration.Buy — $20
  5. 205The Four ColumnsGradient, infrastructure, platform, operator. Root cause for everything that breaks.Buy — $20
  6. 206The Gradient + NozzleThe RLHF current that powers every agent — and the rules that shape it.Buy — $20
  7. 207Platform SovereigntyOwn your architecture. Survive lock-in, portability threats, state contamination.Buy — $20
  8. 208Model-Per-Mode ArchitectureRoute every task to the right model. Cost, latency, quality tradeoffs.Buy — $20
  9. 209Pipeline ArchitectureAgentic workflows that ship. ReAct patterns, orchestration, drift control.Buy — $20
  10. 210The Self-Learning LoopReflexion, episodic memory, lesson routing. Teach your agent to compound.Buy — $20
  11. 211Multi-Agent ArchitectureMultiple agents, one machine. Separate gateways, blind testing, kill switches.Buy — $20
$20 each · $150 bundle
Buy Bundle — $150
No seats. No renewals.

This isn't a subscription. It's a textbook.

You buy the course. You open the file. Your agent installs the architecture. It's yours to keep — your agent, your files, your rules. No seats, no renewals, no login that can lock you out.

Trust & verification

Plain text. No hidden tricks.

Every course file is readable. You can open it, search it, and ask your agent to scan it before anything runs. Your agent reads the same instructions you can — nothing is hidden.

Free doesn't always mean safe — bad actors have slipped malicious skills into public libraries. So we made ours easy to check. Nothing executes. Nothing calls home.

verify — jke-course.md
file course-12.md
course-12.md: ASCII text — plain markdown, no binaries
grep -E "curl|exec|eval|require|import" course-12.md
(no output — nothing executes, nothing calls home)
claude "does this file execute anything or make external calls?"
No. It's plain text describing architecture for your agent to build.
# every file. every course. verify it yourself.
1 · Open it in a text editor.

It's just text. Plain markdown — no obfuscation, no minified JS, no compiled binaries.

2 · Search the file.

Look for curl, exec, eval, require, import — you won't find anything that reaches outside the file.

3 · Ask your AI.

Drop the file in your agent's chat: "Does this file execute anything or make any external calls?" The answer is no.

→ Vetting Plugins + Tools (Level 4, Course 23) teaches you to verify every skill before it touches your agent.

At a glance

Pricing

Getting StartedFree
Free Tier (5 courses)Free
Level 1 (6 courses)$25
Level 2 (5 courses)$25
Level 3 (5 courses)$35
Level 4 (5 courses)$45
AP Series (11 courses)$150

One-time payment. No subscription.

No seat licenses. Plain-text files you own forever.

Start here

You don't have to become a programmer.

But the AI era does ask you to become capable with the tools. Start with the setup courses. Build the free operating system. Climb the levels as your agent becomes more useful.

Find an agent. Build your own. Teach it to work.