Gemini CLI Setup

Gemini CLI is Google's AI assistant. You can authenticate using your Google account, an API key, or Vertex AI.

Install and Configure

Open a terminal (Terminal on macOS/Linux, PowerShell on Windows) and run the following commands.

Install Gemini CLI:

npm install -g @google/gemini-cli

Find the installation path:

::: code-group

which gemini
# Example output: /usr/local/bin/gemini

where.exe gemini
# Example output: C:\Users\Username\AppData\Roaming\npm\gemini.cmd

:::

Open Settings → Agent Client and set the Gemini CLI path to the path found above.
Ensure Arguments contains --experimental-acp (this is set by default).

Authentication

Choose one of the following methods:

If you have a Google account and prefer not to use an API key, you can log in directly.

Run Gemini CLI in your terminal and choose "Login with Google":

gemini

Follow the browser authentication flow.
In Settings → Agent Client, leave the API key field empty.

::: tip If you have a Gemini Code Assist License from your organization, add GOOGLE_CLOUD_PROJECT=YOUR_PROJECT_ID in the Environment variables field. :::

Option B: Gemini API Key

If you prefer to use an API key for authentication:

Get your API key from Google AI Studio
Enter the API key in Settings → Agent Client → Gemini CLI → API key

Option C: Vertex AI

If you are using Vertex AI for enterprise workloads:

In Settings → Agent Client → Gemini CLI → Environment variables, add:

GOOGLE_API_KEY=YOUR_API_KEY
GOOGLE_GENAI_USE_VERTEXAI=true

Leave the API key field empty (use Environment variables instead).

::: tip Gemini CLI natively supports ACP, so no additional adapter is required. :::

Verify Setup

Click the robot icon in the ribbon or use the command palette: "Open agent chat"
Switch to Gemini CLI from the agent dropdown in the chat header
Try sending a message to verify the connection

Having issues? See Troubleshooting.

2.2 KiB Raw Blame History