SALE AI-Academy

Claude Code from scratch, at your own pace

16 levels, 97 lectures, 200+ tasks, real projects — from your first Claude Code run to restoring production after an incident.

Sale price
Checking your price…
Price holds for another
00 d
00 h
00 m
00 s
What this course is

The first AI Academy course

AI Academy is new from CodeGym: we teach the AI skills in highest demand, so you stay ahead of the market instead of chasing it. This is the academy’s first course — about Claude Code, Anthropic’s AI agent that is fast becoming the way into any digital profession, software development above all.

TommyOliverTony Tommy · Oliver · Tony · and others

This isn’t a lecture library — it’s a story. You’re the main character: an ordinary person learning from scratch to build software alongside AI, guided by an experienced mentor. Then clients turn up with requirements of their own, and decisions only you can make.

An agent, not a chatbot

Anyone can ask a model a question — that’s just googling. Few can make an agent work like an agent: it reads the project on its own, edits files, runs commands and comes back with finished work. That’s the difference: people won’t be replaced by AI, but by whoever has learned to direct it.

Start › your working setup

The first steps run right in the browser — nothing to install. After that a single installer sets up your IDE, our task plugin and Claude Code for you — already with our key.

No separate Claude Code subscription to buy — it’s already included.

Who it’s for

No experience needed — the course starts from zero

Each level adds one step. If you have experience, you’ll simply reach the advanced scenarios sooner.

01

Just starting out

A fast route into tech and a solid salary from the start — real projects from day one, not a year of theory.

02

Already a developer

You grow faster, earn more, and become the person who uses AI rather than the one AI replaces.

03

You don’t write code

Any digital profession: first you automate your routine tasks, then you do the work of a whole team.

Curriculum

16 levels:
from first command to production

97+ lectures, 200+ hands-on tasks, 16 tests and extra material for every level — for when you want to dig deeper.

curriculum — claude-code-self
$./course --levels
01 Claude Code: an agent, not a chat
02 Claude Code in your IDE
03 Slash commands and session rollback
04 Skills: ready-made agent abilities
05 Plugins: bundles of capabilities
06 Subagents: a second worker
07 MCP: a bridge to live data
08 Hooks: event-driven automation
09 An autonomous agent: Playwright and /goal
10 CLAUDE.md: project context and memory
11 What the agent costs: /model, tokens and /context
12 Plan mode and permissions
13 Change control: git and /diff
14 Prove you fixed it: logs, tests and /model
15 Reviewing changes: /diff and scope
16 Context: /compact and plan mode
What’s inside

The whole workflow with an agent, not a bag of tricks

16 levels and 97 lectures, told as a comic — laid out in the order the work with an agent actually happens:

sessions slash commands skills subagents MCP hooks plan mode git workflow
01 First game
02 First client
03 Tetris
04 Snake, but neon
05 A site that lives online
06 A second pair of eyes
07 Earthquake radar
08 A bot that runs itself
09 Chess against AI

Plus plenty of other projects — we’ll leave those as a surprise. And extra material on development basics, if you need something to lean on along the way.

Honestly

“Is it actually worth it?”

An honest answer isn’t “of course it is” — it’s what’s actually inside:

01

Claude Code is ready to go

Access to Claude Code with CodeGym’s custom models is part of the subscription — nothing to connect or pay for separately. The agent is tuned for the course: its own answer style, context via CLAUDE.md, custom skills, hooks and permissions.

02

200+ tasks — a simulator for real work

Every task rehearses the real process: you write the spec, break it into steps, run the agent in plan mode, review the diff, fix bugs from logs and stack traces, work with branches and rollbacks, wire up MCP and configure hooks.

03

Instant feedback, not a day later

Submit a solution and you immediately get a report on which requirements passed and which didn’t — no guessing what went wrong. It’s checked on two counts: whether it formally works, and whether the logic of your solution holds up.

04

If your solution could be better, the AI mentor says so

The AI mentor gives specifics: how to word the spec more precisely, where to narrow the scope, how to manage context, when to move a repeated step into a skill, how to build a safe git flow.

Deadline

The price goes up once

The sale price holds until the sale ends, then the regular price returns. Nothing in between.

4.9 out of 5 2,000,000+ have studied at CodeGym 35,000+ now working in tech
© CodeGym, 2026 Claude Code Premium