Get Ready for the Course

Accounts to open and tools to install before Day 1 — takes about 30 minutes

// PRE-COURSE SETUP · 课前准备

GitHub + Copilot

GitHub is where your team's code lives — it's a deliverable. Copilot is GitHub's AI pair programmer, built into VS Code. It completes code, explains unfamiliar functions, and writes boilerplate so you can focus on the interesting parts.

Create a free GitHub account at github.com
Apply for the GitHub Education Pack at education.github.com/pack — use your @shanghaitech.edu.cn email. Approval takes 1–3 days and gives you Copilot for free.
Install GitHub Copilot as a VS Code extension (step 5 — set up VS Code first if needed)

→ github.com → Education Pack

Apply for the Education Pack as early as possible — approval is not instant.

OpenRouter

OpenRouter is a single API gateway to 300+ AI models — Qwen, DeepSeek, Claude, and others. Your ESP32 will call OpenRouter to get your robot's AI responses. You use one API key, one endpoint, and can swap models with a one-line change.

Create a free account at openrouter.ai
Go to Settings → API Keys and create a key — save it somewhere safe
Add $5–10 of credit under Settings → Credits — this will last the entire course

→ openrouter.ai → Browse models

$5 is genuinely enough. Each LLM call costs fractions of a cent at the models we use.

Blender

Blender is the 3D modeling tool you'll use to design your robot's enclosure — the physical body that makes it a product, not just a circuit on a table. It's free, open-source, and runs on any computer. You don't need to know it yet — we'll learn the basics together.

Download and install Blender from blender.org/download — get the latest stable release
Optional head start: watch the first 2 episodes of the Blender Guru donut tutorial — it covers the core interface in ~30 minutes

→ blender.org/download

Free forever. No account needed.

Claude Pro

Claude Pro gives you access to Claude Code — Anthropic's agentic AI coding tool that runs in your terminal. It understands your entire codebase, writes firmware, explains ESP-IDF errors, and can refactor across multiple files at once. It's the fastest way to go from "I have no idea what this error means" to "it's working."

Sign up at claude.ai — use any email
Subscribe to Claude Pro ($20/month) — check if your institution has a discount or arrangement
Install Claude Code in your terminal: npm install -g @anthropic-ai/claude-code

→ claude.ai → Claude Code docs

Requires Node.js. If you don't have it, install from nodejs.org first.

VS Code + PlatformIO + ESP-IDF

VS Code is the editor. PlatformIO is the extension that handles ESP32 board config, library management, and flashing. ESP-IDF is Espressif's official development framework that PlatformIO uses under the hood. Together they turn your laptop into a proper embedded development workstation.

Download and install VS Code from code.visualstudio.com
Open VS Code → Extensions (⌘⇧X) → search PlatformIO IDE → Install
PlatformIO will install ESP-IDF automatically when you first open an ESP32 project
While you're in Extensions, also install GitHub Copilot (from step 1)

→ VS Code → PlatformIO for VS Code → ESP-IDF docs

All free. The PlatformIO first-time install downloads ~500 MB of toolchains — do it on Wi-Fi.

Already done? Try a hello world with Claude Code.

All five tools installed? Let Claude Code write your first firmware. Here’s how:

1. Create a PlatformIO project in VS Code

Open VS Code → PlatformIO icon → New Project → Board: Espressif ESP32 Dev Module → Framework: Arduino → Finish.

2. Open Claude Code in the project folder

cd your-project-folder
claude

3. Tell Claude Code what you want

Write an Arduino sketch for ESP32 that blinks the built-in LED on pin 2
every 500ms. Put it in src/main.cpp. Then tell me the PlatformIO command
to build and upload it.

Claude Code writes the code, places it in the right file, and gives you the exact command to flash it. If something fails, paste the error back and ask it to fix it — that loop is the core skill you’ll use all course.

Don't panic. We're all figuring this out.

Some of these tools didn't exist two years ago. One or two will probably change before the course starts. The documentation will have gaps. The AI will occasionally hallucinate something spectacular. This is not a sign you're doing it wrong — it's a sign you're at the frontier.

Your instructors are learning too — we just have a head start of a few months. The best thing you can do is show up curious, set up the tools above, and be ready to break things on Day 1.

Get ready to break things on Day 1.