Home/ Front Matter/0.4 Required vs optional tools (AI Studio, editor, Git, runtime)

0.4 Required vs optional tools (AI Studio, editor, Git, runtime)

Overview and links for this section of the guide.

On this page

Minimum required setup

You can do a surprising amount with just a browser and Google AI Studio. But to actually ship software (run code, debug, iterate), you need a small local toolchain.

  • Google AI Studio access: a Google account with AI Studio available in your region.
  • A modern browser: Chrome, Edge, Firefox, or Safari.
  • A code editor: anything you can comfortably edit multiple files in.
  • A terminal: to run programs, tests, and scripts.
  • A runtime: pick one primary runtime to start (Node.js or Python are the most common in this guide).
Pick one lane first

Choose either Node.js (JavaScript/TypeScript) or Python as your “default runtime” for early projects. Switching later is easy; switching every day is not.

These tools dramatically improve your iteration speed and reduce the “AI wrote it, now I’m scared to touch it” problem.

  • VS Code (or another IDE you already like)
  • Extensions: language support, formatter, linter, Git integration
  • Node.js: use an LTS version; consider a version manager (e.g. nvm)
  • Python: use a modern 3.x version; consider an environment manager (e.g. pyenv + venv)
  • JavaScript/TypeScript: Prettier + ESLint
  • Python: Ruff (lint) + Black (format) or Ruff-only

When you vibe code, tests are how you keep speed over time.

  • JavaScript/TypeScript: Vitest or Jest
  • Python: Pytest
  • Baseline: a single “smoke test” that proves the app still runs

Git is optional for reading, but strongly recommended for building.

  • Why it matters: you can experiment freely and roll back when the model takes a wrong turn.
  • How to use it in this guide: small commits at checkpoints and ship points.
A high-signal workflow

Ask the model for diff-sized changes, then review and commit. Small diffs + Git history = fearless iteration.

Optional tools (add when needed)

These are useful, but you don’t need them on day one. Add them when your project forces the requirement.

  • Docker: when you need reproducible environments or multiple services.
  • Makefile / task runner: when commands start getting long or repetitive.
  • HTTP client: Insomnia/Postman/curl for testing APIs quickly.
  • Database tooling: SQLite browser, Postgres client, migrations framework.
  • CI: GitHub Actions (or similar) once you have tests worth running on every change.
  • Secrets manager: once you have real credentials and multiple environments.
Avoid tool pile-up

More tools don’t automatically mean more speed. Add tooling in response to a specific pain (reproducibility, regressions, deployment), not because it’s “best practice.”

Safe defaults for beginners

If you just want a working baseline with minimal choices, this is a solid default stack:

  • Editor: VS Code
  • Runtime: Node.js LTS
  • Formatting: Prettier
  • Testing: Vitest (or a simple smoke script)
  • Version control: Git (local only is fine)

Quick setup checklist

  • AI Studio: confirm you can open AI Studio and start a prompt session.
  • Editor: open a folder, create a file, run a formatter.
  • Runtime: verify you can run node --version or python --version.
  • Project loop: create a tiny “hello world” app and run it from the terminal.
  • Git (optional but recommended): initialize a repo and make a first commit at your first checkpoint.
You don’t need everything installed

The only requirement is that you can do the loop: prompt → code → run → verify → refine. Everything else is there to make that loop faster and safer.

Where to go next