Week 1

Amplifier, not a replacement

How an agent works

Setting up Claude Code

First project and CLAUDE.md

Figma → code via MCP

Practice

Assignment

Back to Track

Prerequisites

• Claude Code Account. It requires a paid Claude subscription (Pro or higher) or Console (API) access — the free Claude.ai plan does not include Claude Code.
• VS Code installed.

Watch: What is Claude code?

https://www.youtube.com/watch?v=fl1DSmwQKKY

Two ways to run Claude Code

Claude Code comes in two forms, and both are first-class:

• Desktop app — a graphical interface: you point it at a project folder and work with the agent directly, no terminal needed. This is the main way we'll work in this track.
• CLI — the same agent running in the terminal. Some developers prefer living in the terminal; and either way you'll use the CLI occasionally (some things, like installing plugins, are done by command).

Pick Desktop as your base; set up the CLI too — it takes two minutes and you'll need it.

Option A — Desktop app (our base)

1. Download and install the Claude app: https://claude.com/download
2. Sign in with your Claude account.
3. Open the Code section and point it at a project folder.

That's it — you're talking to the agent about your project. The folder you select is what the agent sees as its working context.

Option B — CLI (set it up too)

The terminal is still part of a developer's life (Git, running your project's commands), so get minimally comfortable: open the terminal, see where you are (pwd), list files (ls), move around (cd folder, cd ..), make folders (mkdir). If that's new, work through Anthropic's official beginner terminal guide: https://code.claude.com/docs/en/terminal-guide — if you completed the Core program, skim and move on.

Install with the native installer (it keeps itself updated automatically):

macOS / Linux / WSL:

curl -fsSL <https://claude.ai/install.sh> | bash

Windows (PowerShell):

irm <https://claude.ai/install.ps1> | iex

On native Windows, installing Git for Windows (https://git-scm.com/downloads/win) is recommended — it gives Claude Code a proper Bash shell.

Verify:

claude --version

A version number = success. For a deeper health check: claude doctor. If anything fails: https://code.claude.com/docs/en/troubleshoot-install

First launch: run claude from any folder — on first run it opens a browser to log in. When run from a project folder, the agent gets that folder as context. Ctrl+C twice to exit.

VS Code

Install the official Claude Code extension from the VS Code marketplace and sign in with the same account. It gives you the agent right next to your code: open the project folder in VS Code and start Claude from the Claude panel (or run claude in the terminal panel from the project root).

Whichever entry point you use — Desktop, VS Code, or terminal — the rule is the same: the agent sees the folder it is started in. Always start it from the project root.

One habit to start now: role-chats

Different chats for different roles: one chat that codes, a separate one that reviews, another for design questions. Pin the important ones.

Why it works: each chat keeps a clean context for its task (context is limited and degrades as it fills), and a separate review chat hasn't "seen" how the code was written — it looks at the result with fresh eyes instead of defending its own decisions.

For this week's project one chat is enough. Split into roles once a project grows. The automated version of this idea — subagents — comes in Week 5.

The HackYourFuture curriculum is licensed under CC BY-NC-SA 4.0 *https://hackyourfuture.net/*

Built with ❤️ by the HackYourFuture community · Thank you, contributors

Found a mistake or have a suggestion? Let us know in the feedback form.