Skip to content

How to build an app with Claude Code

By Thijs Verreck · Published Nov 2, 2026
Build an app with Claude Code, step by step: install it, describe the app, see it run, and commit it. Quoted from Anthropic's docs, read 31 Oct 2026.

Claude Code is Anthropic's coding agent. Its documentation describes it as "an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools." This guide takes you from an empty folder to a running app you can commit. The steps follow Anthropic's docs as of 31 October 2026.

What you need

  • A Claude plan or a Console account. Anthropic's overview says "most surfaces require a Claude subscription or Anthropic Console account." For prices, see AI app development cost.
  • A terminal, or one of the other places Claude Code runs. The docs list "the terminal, IDE extensions, a desktop app, and the web." This guide uses the terminal.
  • An idea for the app, in a sentence or two.

1. Install Claude Code

On macOS, Linux or WSL, Anthropic's install command is:

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

Windows has its own commands on the same page. When the installer finishes, open a new terminal and run claude --version. The docs say "a working installation prints a version number."

2. Start it in a project folder

Make a folder for the app, go into it and start Claude Code:

Bash
mkdir my-app && cd my-app
claude

Anthropic's docs say "Claude Code prompts you to log in on first use."

3. Describe the app

Say what the app is for, who uses it, and what the first screen shows. Anthropic's docs put it this way: "Describe what you want in plain language. Claude Code plans the approach, writes the code across multiple files, and verifies it works." Start small. One screen that works beats ten that half work.

4. Write down your rules

Add a CLAUDE.md file to the project. Per the docs, it's "a markdown file you add to your project root that Claude Code reads at the start of every session." Put in it what should stay true on every change, such as the framework, the styling and what not to touch.

5. See it running, and steer

Ask Claude Code how to start the app, then open it and use it. Say what's wrong and what should change, one thing at a time. That's the loop from the AI app development guide.

To build where your team can see it, connect Claude Code to Prototyper over MCP with one command:

Terminal
claude mcp add --transport http prototyper https://www.prototyper.co/api/mcp

Claude Code then writes the app's files in your Prototyper workspace, and the app runs on the shared canvas next to the team's sketches. The connect guide has the details.

6. Commit it

Per the docs, "Claude Code works directly with git. It stages changes, writes commit messages, creates branches, and opens pull requests." Ask it to commit when a step works, so you can always go back.

Source: Anthropic, Claude Code overview, read 31 October 2026.

Frequently asked questions

Do I need to know how to code?
No. You need to know what you want and to say what's wrong when you see it. Reading code helps when something breaks, and for payments, private data and scale you'll want a developer.
Is Claude Code free?
Anthropic's docs say most surfaces need a Claude subscription or an Anthropic Console account. The cost page lists the plans, with prices as of 30 Oct 2026.
Can I use Claude Code with Prototyper?
Yes. Add Prototyper as an MCP server with the command in step 5, and Claude Code builds in your workspace on the plan you already pay for.