Set Up Your AI Coding Tools

Before you install anything: pick your path

If you haven't read the overview, do that first — it decides which of the three paths (wrapped web app, cross-platform, native Android) you're actually building toward. All four tools below can build any of the three; the choice of tool is about which company's product you're already comfortable with, not which path you're taking.

Claude Code (Anthropic)

The most established of the four for this kind of work, runs in a terminal or inside VS Code.

1. Install VS Code if you don't have an editor already (code.visualstudio.com).
2. Install Claude Code with the one-line command from code.claude.com/docs — there's a Windows line and a macOS/Linux line.
3. Make an empty folder for your project (e.g. my-export-app). In VS Code: File → Open Folder → pick it.
4. Open the terminal panel, type claude, press Enter, and trust the folder.
5. Type your first request in plain English — see the prompts in the next guide.

Needs a Claude subscription (or API billing) once you go past a light free allowance.

Codex (OpenAI)

OpenAI's equivalent coding agent, usable from the terminal or inside ChatGPT.

1. If you already use ChatGPT, Codex may already be available to you in the same account — check the ChatGPT sidebar for a "Codex" entry, or install the codex CLI following OpenAI's current setup docs.
2. Point it at an empty project folder the same way as Claude Code: open the folder, run the tool inside it, describe your first request.
3. A reasonable pick if you'd rather keep everything under one subscription you already pay for.

Gemini CLI (Google)

Google's terminal coding agent, with a genuinely usable free tier — the easiest of the four to just try before spending anything.

1. Install via npm install -g @google/gemini-cli (Node.js must already be installed), or follow Google's current install docs if the package name has changed.
2. Run gemini inside your empty project folder and sign in with a Google account.
3. Describe your first request the same way as the others.

Its natural advantage: if you go the Firebase route later (real-time data, image storage, simple accounts, mentioned in the prompts guide), Gemini integrates most naturally with it since both are Google products.

Antigravity (Google)

Google's newer agentic development environment, built around Gemini models, aimed specifically at "describe the app, get a working build" style development rather than a plain terminal session.

1. Download and install it from Google's current distribution page for Antigravity.
2. Open (or create) your project folder inside it the same way as any code editor.
3. Describe what you want to build in its prompt panel — the same prompts in the next guide work here too, since the underlying task (an agent writing and running code in your folder) is the same shape as the other three tools.

Being newer, expect its exact interface and setup steps to keep changing faster than the other three — check Google's current documentation if something here is out of date.

Which one should you actually use?

None of the four is a strict upgrade over the others — they converge on the same idea (an agent that writes code for you in a folder) faster than any one company's marketing suggests. The practical rule: start with whichever company's product you already pay for. If you're not already paying for any of them, Gemini CLI's free tier is the lowest-friction way to try the whole idea before committing money. Switching tools later costs you a re-explained project brief, not lost work — the code stays in your folder regardless of which tool wrote it.

Next: Copy-paste prompts to actually build your app →