Give Gemini CLI control of your browser

The problem

• Gemini CLI has no persistent browser and cannot reach logged-in pages.
• Fetching a URL is not the same as operating an app.

What Vibe MCP changes

• Gemini drives your real Chrome, with your sessions.
• Same tool surface as every other Vibe MCP client.

Install in 3 steps

1. 1Install the Vibe extension in Chrome

   Add Vibe AI Browser from the Chrome Web Store (Chrome, Brave, Edge, or any Chromium browser). This is the piece that lets an agent drive the browser you are already logged into — no second browser profile, no re-authentication.

2. 2Add Vibe to ~/.gemini/settings.json

   Add the "vibe" MCP server to your Gemini CLI settings file and restart the CLI.

   ~/.gemini/settings.jsonjson

   {
     "mcpServers": {
       "vibe": {
         "command": "npx",
         "args": ["-y", "@vibebrowser/mcp"]
       }
     }
   }

3. 3Connect the extension and run your first command

   Click the Vibe icon, open Settings, and enable "MCP External Control" until the status reads Connected. Then ask your agent to open a page and take a snapshot — it should return the content of your real tab.

Your relay URL / extension UUID grants live control of your browser session. Treat it like a password: never commit it or paste it into a shared chat.

Tools Gemini CLI gets

How it compares

Gemini CLI browser control — FAQ

Does Gemini CLI support MCP?

Yes. Gemini CLI loads MCP servers from ~/.gemini/settings.json. Adding the Vibe server there gives Gemini browser tools backed by your real Chrome.

Can Gemini read pages behind a login?

Yes, because it acts inside your existing browser session rather than a fresh automation profile.

Works with your other agents too

The same MCP server powers every client. Multiple agents can drive the same browser at once.