Connect your app

Gemini CLI

Add Ollive to Gemini CLI from your terminal.

Before you start

Install Gemini CLI on a computer with a web browser: the first sign-in opens one.

You also need an Ollive login with access to the workspace you want to use. There’s no API key to create.

Run this in your terminal:

Terminal

gemini mcp add --transport http --scope user ollive https://mcp.ollive.ai/mcp

With --scope user, Ollive is available in every project. Leave it out to add Ollive to the current project only.

Or add it to ~/.gemini/settings.json (every project) or .gemini/settings.json (one project) yourself:

settings.json

{  "mcpServers": {    "ollive": {      "httpUrl": "https://mcp.ollive.ai/mcp"    }  }}

Start Gemini CLI and run this. Your browser opens Ollive.
Gemini CLI
```
/mcp auth ollive
```
Sign in to Ollive if you’re asked. Check the app and the workspace, then select Allow. When the page says Gemini CLI is connected, close the tab and go back to Gemini CLI.

Run /mcp. Ollive shows as connected, with its tools. Then ask:

AskWhich of my agents does Ollive monitor?

Ollive is listed but not connected: Run /mcp auth ollive to sign in again.
Sign-in doesn’t finish: Gemini CLI finishes sign-in through a browser on the same computer. If you work over SSH or in a container, sign in where a browser can open.
Ollive doesn’t connect at all: Check that your settings use httpUrl, not url.
You want a different workspace: Run /mcp auth ollive and choose the other workspace.

Still stuck? Contact support with the app’s name and the message you see.