learn python like a game

Learn to code the way
real work actually looks.

An AI agent grades your code, dials in the difficulty, and walks you through the path: syntax → data structures → OOP → production. Real editor, terminal, and git — in the browser, no install.

Closed beta phase · seats given out in cohorts · signup takes 4 steps

$ pytest task_042.py
4/4 tests passed · +25 XP · rank: Adept
> next task: decorators
HP
520 / 700
XP
+25 today

One path, four layers woven together

The mentor doesn't teach topics in isolation — it weaves them into one continuous path, shaped around your goals.

01 / fundamentals

Programming fundamentals

Syntax, data structures, algorithms, paradigms and Python idioms — the foundation everything else stands on.

02 / specializations

Specializations

Web, data science, ML, deep learning, AI agents/LLMs, game dev, GUI apps, security — an open catalog, growing with more paths.

03 / process

The engineering process

Project structure, git flow, writing tests (unit, integration, TDD, mocking), CI/CD, debugging, code review — what the job actually looks like.

04 / emergent

Emergent topics

Threads that surface as you talk with the mentor. Mention a banking API? The agent adds nodes on authorization and idempotency.

Eight mechanics, one goal: never bored, never stuck

1

Instant feedback

Mentor grading and test results in under 30 seconds from submission.

2

Progress always visible

Skill map, XP bar, and rank — progress is never a guess.

3

Difficulty that fits

The agent holds success around 70–80% — the zone where you learn fastest, without hitting a wall or getting bored.

4

Competition

Global and weekly leaderboards, rank leagues — for when progress alone isn't enough.

5

Habit

Daily streaks and an agent-generated daily quest, calibrated to your level.

6

No end to the game

Specialization paths and an "endless mastery" mode — there's always another level down.

7

Process realism

The full engineering cycle: project structure → feature branch → PR → agent code review → CI/CD → merge. Skills "like at work," not just katas.

8

Decision economy

HP as hint currency — you negotiate the price of a hint with the agent. Every bit of help is a conscious cost.

HP is life. XP is proof you're better than yesterday.

Amber (HP) is reward and life — the currency you negotiate with the agent for hints. Violet (XP) is the agent and intelligence — the measure of your real progress. Theory never costs HP — you only pay for hints on a specific task.

70–80%
target success rate on tasks — the zone of proximal development, kept in range by the agent in real time
<15 min
typical time to your first completed task and first points, from registration
<30 s
time to grade a submitted solution — tests, lint, and mentor feedback
Novice
0 XP
Apprentice
500 XP
Adept
1,500 XP
Specialist
4,000 XP
Expert
9,000 XP
Master
20,000 XP
Grandmaster
top 1% of season

A real workshop, not a simulator

Zero install, zero environment setup. Open the browser and write code the way you will at work.

Editor, terminal, and git — from minute one

Real Python running directly in the browser, real git commands, a real terminal. Your progress saves automatically between sessions.

Deterministic grading, not "I think so"

Tasks are checked by tests and static analysis — not subjective opinion. You know the result immediately, no queue to wait in.

Disagree with a grade? Contest it.

Every task can be challenged in the mentor chat — an independent second review takes your argument into account.

Documentation at hand, not in another tab

Built-in Python documentation search with cited sources — used by both you and the mentor before it offers a hint.

task_042.py
test_task_042.py
# task: safely fetch a nested value
def safe_get(data: dict, *keys, default=None):
    for key in keys:
        try:
            data = data[key]
        except (KeyError, TypeError):
            return default
    return data

# $ pytest -q
# ✓ 4 passed in 0.21s

Fair grading, zero black box

The result is never just an opinion from a language model. Objective signals — tests (including hidden ones), mutation tests, lint, security checks — weigh at least 70% of the grade. The rest is a quality rubric, always with reasoning visible to you.

tests, lint, security — ≥70%quality rubric — ≤30%
Anti-cheat philosophy: disarm, don't chase.

What that means in practice

One appeal per task — an independent second review, without access to the first grade.

The leaderboard only counts honestly-earned solutions — learning always stays available, regardless of ranking status.

A permanent account ban is always a human decision — never an automated one.

Wherever you're starting from, the path adapts to you

starting from zero

You're starting to code

No experience? An initial assessment picks tasks at your level, and theory explains every new concept before you hit it.

career change

You're switching careers into tech

You learn not just syntax, but process: git, tests, code review — exactly what employers look for.

leveling up

You already work with code

Go deeper into a specific specialization — data science, AI engineering, security — at a pace and format you choose.

teams

You're building a team

A plan for companies: a consistent dev environment for onboarding, and skill-development reports for your team. Get in touch.

We're building this in public, step by step

No infrastructure jargon — just what actually unlocks for you at each phase.

0
now · closed beta

The first loop: task → grade → reward

Initial assessment, adaptive Python tasks (katas and debugging), theory calibrated to your level, points and your first rank. The task bank grows from day one of play.

1
full gameplay

The complete skill tree

Every specialization unlocked, hint-price negotiation, free-form chats with the mentor, ranked seasons and leagues, a formal grade-appeal process.

2
process mode + teams

A real engineering cycle and team plans

A full project repository, pull requests, agent code review acting as a senior engineer, CI/CD pipelines — plus licensing plans for teams with a consistent dev environment and development reports.

3
scale

More languages, a mature task bank

Interface expansion into more languages, full development reports for companies, a mature task ecosystem calibrated on real gameplay data.

Rule for opening the next phase: never a gut call. We move to the next stage only once closed-beta data shows the motivation loop actually keeps players around.

Closed beta — sign up

Before you play the game, play this form. Four steps, just like the real initial assessment — they help us place you in the right cohort with the right first tasks.

Discovering your level: 1/4
We'll send your invite to this address once your cohort opens.

✓ application received

You're on the list.

We'll email you once your closed-beta cohort opens. Until then — no spam, and no HP will vanish, because you don't have any yet.

Please fill in the required fields (*) before continuing.

Questions you'll ask anyway

Do I need to install anything?

No. The editor, terminal, and Python all run directly in the browser. You just need a browser that supports WebAssembly and an internet connection.

How much does the closed beta cost?

Nothing. The beta is free. Subscription and company plans will appear at full launch — beta participants will hear about them first.

Do I need prior experience?

No. The initial assessment picks tasks matched to your real level — from zero to professional work with code.

Do points, XP, and HP have monetary value?

No. They're gameplay elements — not a means of payment, and not exchangeable for money or anything else. Details in the Terms.

What happens to my data and chats with the agent?

You have full visibility into everything the system knows about you, self-service data export, and you can delete your account anytime. Full details in the Privacy Policy.