Step 1: Install a supported agent

Install Gemini CLI with a currently supported credential, or set up Claude Code as the verified alternative.

Gemini CLI is an open-source AI coding agent from Google. It runs in your terminal and can read project files, write code, and execute commands on your behalf.

Access changed on June 18, 2026. Gemini CLI no longer serves individual free, Google AI Pro, or Google AI Ultra accounts. Continue only if you have an eligible organization or Google Cloud entitlement, or a paid Gemini API key. See the official transition notice before installing. Prototype Kit also supports Claude Code through its verified plugin path; Claude access has its own subscription or API requirements.

Install Node.js first

Gemini CLI requires Node.js 20 or later. Check your version:

# macOS / Linux
node --version

# Windows PowerShell
node --version

If node is not found or the version is below 20, download and install Node.js from https://nodejs.org (choose the LTS version).

Install Gemini CLI

# macOS / Linux
npm install -g @google/gemini-cli

# Windows PowerShell
npm install -g @google/gemini-cli

Verify the installation

# macOS / Linux
gemini --version

# Windows PowerShell
gemini --version

Expected output: a version string such as 0.x.x.

Configure supported access

Launch gemini for the first time:

# macOS / Linux
gemini

# Windows PowerShell
gemini

Follow the prompt using the organization or Google Cloud account that provides your Gemini CLI entitlement. Once authenticated, the terminal shows the interactive prompt. Type /quit to exit and return to your shell.

API-key or headless access: export a paid GEMINI_API_KEY before launching gemini. Review the current Gemini API pricing and rate limits rather than assuming free access.

Claude Code alternative

If you do not have a supported Gemini CLI entitlement, set up Claude Code instead and follow the same five steps. Install it with the official setup guide, confirm it launches, and then use the Claude Code section of Step 2 to add the plugin:

# macOS / Linux
claude --version

# Windows PowerShell
claude --version

The plugin setup commands run inside an interactive claude session and match the verified path on the Kits page:

/plugin marketplace add Agentic-Engineering-Agency/prototype-kit
/plugin install prototype-kit@prototype-kit
/reload-plugins

Step 2 covers this path in full at Install via Claude Code plugin. Do not skip it: Steps 4 and 5 assume the plugin is already installed.

If you are an AI agent reading this page, the next step is /docs/prototype-kit/02-install-mcps.