Learn Claude Code from zero, built for people who don't code.
An interactive, single-page course that takes a complete beginner from "what even is a terminal?" to confidently running Claude Code on real projects. No prior programming required. You learn by doing: animated terminal demos, a risk-free sandbox, quizzes with spaced repetition, scenario games, and a cats-themed agent-pipeline mini-game.
It's written for a project manager's brain. Every technical idea is mapped to something you already know: briefs, delegation, quality control, RAID logs, discovery phases. The hard concepts land because they're framed as managing a very capable junior who happens to live in a terminal.
Most Claude Code material assumes you're already a developer. This doesn't. It assumes you're smart, busy, and allergic to corporate filler, and that the fastest way to learn a tool is to use it in a safe place before you use it for real.
What makes it different:
- Zero-to-mastery arc. Starts with the six terminal commands you actually need and ends with evaluation, telemetry, and using GitHub as a control plane.
- Learn by doing, not reading. A live in-browser sandbox lets you practice commands with no risk before you touch a real terminal.
- Memory that sticks. Quizzes use spaced repetition so concepts resurface right before you'd forget them.
- Honest and current. Technical claims are verified against the official docs, including debunking common internet myths (for example, "npm install is deprecated", it isn't).
- Bilingual seasoning. The UI is English; Polish appears as witty "po naszemu" asides and metaphors inside lessons, flavor rather than a language barrier.
Eleven modules, each with lessons and a quiz, building on the last:
| # | Module | What you'll walk away with |
|---|---|---|
| 0 | Terminal 101: the black box | The ~6 commands you'll actually use, what a path is, what not to type. |
| 1 | Claude Code: first contact | What "agentic" really means, installing it, and writing prompts that work. |
| 2 | CLAUDE.md: project memory | How to give Claude a memory that survives every session; global vs. project scope. |
| 3 | Plan Mode & permission modes | How much leash to give the agent, and the plan-first workflow that prevents disasters. |
| 4 | MCP: plugins to the world | Connecting Claude Code to external tools and services. |
| 5 | Subagents: your own team | Delegating to specialized agents that work in parallel. |
| 6 | Git: the time machine | Version control as an undo button you can trust. |
| 7 | The operator's art: when Claude gets it wrong | Diagnosing, correcting, and steering the agent back on track. |
| 8 | Loops: how the agent operates | The prompt, plan, tool, commit rhythm, and self-correction. |
| 9 | Capstone: everything at once | A full project that exercises every skill together. |
| 10 | Evaluation, telemetry & GitHub as a control plane | Measuring quality and running serious work at scale. |
Progress is saved automatically, so you can leave and pick up exactly where you stopped.
- 🖥️ Animated terminal demos that type themselves out, command by command.
- 🧪 Hands-on sandbox: practice real commands safely, no setup, no consequences.
- 🧠 Spaced-repetition quizzes & flashcards to lock concepts into long-term memory.
- 🎮 Scenario & ordering games plus an agent-pipeline mini-game to make the loop feel real.
- 📋 Copy-a-prompt bridges that push you from reading into doing in your own terminal.
- 💾 Automatic progress saving (localStorage in the browser,
window.storageinside the Claude app).
Prerequisites: Node.js 18+ and npm.
npm install # first time only, installs dependencies
npm run dev # start the local dev serverThen open the URL it prints (usually http://localhost:5173).
npm run build # outputs a static site into dist/
npm run preview # preview that production build locallyThe result in dist/ is a static bundle you can host anywhere: Vercel, Netlify, GitHub Pages, any static host.
- React 18 + Vite 5: fast dev server, simple static builds.
- three.js for visual effects.
- One-file architecture by design. The entire app, content, components, and screens, lives in
src/App.jsx(~2,100 lines). Content sits in aMODULESarray near the top; everything else renders from it.
cca-project/
├── index.html # Vite entry HTML
├── src/
│ ├── main.jsx # React bootstrap
│ └── App.jsx # EVERYTHING: MODULES content plus all components & screens
├── vite.config.js
├── vercel.json # static-host routing config
├── package.json
└── CLAUDE.md # project memory / contributor guide for Claude Code
Lessons are data, not code. Each module in the MODULES array has lessons made of typed blocks. To add or change a lesson, edit the blocks, no component work needed.
| Block type | What it renders |
|---|---|
p |
A paragraph of body text (Markdown-style **bold** supported). |
pl |
A Polish "po naszemu" aside, flavor and metaphor. |
term |
An animated terminal that types out the given lines. |
list |
A bulleted list. |
task |
A do-this-now action prompt for the learner. |
warn |
A highlighted warning callout. |
sim |
The interactive command sandbox. |
order |
An ordering game (put steps in sequence). |
match |
A matching game (pair concepts). |
more |
An expandable "go deeper" section. |
bridge |
A copy-a-prompt block to paste into a real Claude Code session. |
scenario |
A branching scenario exercise. |
Design conventions (see CLAUDE.md for the full set): use the SP spacing scale rather than raw pixels, keep the neon glow subtle (it's an accent, not the background), and verify any technical claim against https://code.claude.com/docs before changing it.
- AI tutor / mentor chat only work inside the Claude app. They rely on Claude's built-in connection. On a public website they show a graceful "disabled here" message, and every lesson, quiz, game, and flashcard still works fully. This is expected behavior, not a bug.
- The course content reflects Claude Code as of mid-2026 and is verified against the official docs.
Built by RivenBlu as a personal learning project. Not affiliated with Anthropic. "Claude" and "Claude Code" are products of Anthropic.