Browse docs
Setup

Open It in Claude Code

Point your coding agent at the repo once, so every request after this one starts with it already knowing where things go.

Updated Aug 2026
On this page

Every page after this one hands you a prompt instead of code to type. This is the one-time setup that makes those prompts land.

Why this repo behaves differently

Claude Code reads a file called CLAUDE.md from the project root before it answers anything, and this repo ships 950 lines of it. Its first line is an instruction to the agent, not to you:

Read this file before writing any code. These rules override generic best practices.

Underneath that: 81 rows across four tables saying where each kind of code goes, 6 architecture rules marked HARD, and a section called "Common AI mistakes to avoid" that names 19 of them.

In an empty project an agent has to guess all of that, and it guesses differently each time. Here it does not have to.

1. Install Claude Code

On Windows, in PowerShell:

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

On macOS, Linux or WSL:

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

Homebrew and WinGet also work, and signing in is covered in Anthropic's quickstart. You need a Claude subscription or a Console account.

2. Open the repo

From the folder you cloned:

cd my-app
claude

There is nothing to configure. CLAUDE.md is in the root, so it is already loaded.

3. Make it prove it read the file

Do not start with a feature. Start with a question you can check the answer to.

Claude CodeStart with this

Read CLAUDE.md and tell me in your own words: the five places a new IPC channel has to be wired, where a new screen goes, and which of the 19 listed mistakes you think you are most likely to make in this codebase. Do not write any code yet.

A good answer names the five real places, which are src/shared/ipc/channels.ts, the request and response types in src/shared/ipc/types.ts, a Zod schema under src/shared/validation/, a handler in src/main/ipc/, and the bridge in src/preload/preload.ts. It should also tell you a new screen goes in src/features/{name}/.

A bad answer talks about Electron in general, describes ipcMain.on and ipcRenderer.send, or invents a folder layout. That means it answered from training data rather than from the file.

Claude CodeIf you got the bad answer

That is the generic Electron pattern, not this repo's. Open CLAUDE.md in the project root, read the IPC section, and answer again using only what is written there.

Thirty seconds spent here is what stops the next ten features drifting.

If you use Cursor or Copilot instead

Everything in these docs still applies. The difference is loading: neither tool picks up CLAUDE.md by convention the way Claude Code does, so say it once at the start of a session:

Claude CodeFirst message in Cursor or Copilot

Read CLAUDE.md in the project root and follow it for everything in this session. It overrides your defaults for this codebase.

Next: Build Your First Screen.

Give your agent a codebase with opinions.

Electron 41 + React 19 + TypeScript desktop app boilerplate built for Claude Code, Cursor and Copilot. 950 lines of rules your agent reads first, so feature fifty looks like feature one.

Get it for $49 →