CultivX Browser Control

Exposes this browser to the standalone browser-control-mcp server for agent-driven navigation.

CultivX Browser Control lets an AI agent work inside the browser you already use. Most browser automation starts a clean, empty browser and asks you to sign in again. This does the opposite. It connects the browser you are already signed into to an MCP client such as Claude Code, Claude Desktop, Cursor or VS Code, so the agent sees the tabs, sessions and extensions you actually have. WHAT IT DOES • Navigate, open, close and group tabs. • Read the visible text of a page. • Click, type, scroll and select the way you would. • Take screenshots. • Upload a file you choose. HOW IT WORKS The extension connects to a server running on your own computer at 127.0.0.1. That server is part of the free, open-source @cultivx/browser-control-mcp package, which your AI application starts. The extension does not talk to the internet on its own: no analytics, no account, no CultivX server. SETUP 1. Install this extension. 2. Add the server to your MCP client: npx @cultivx/browser-control-mcp --stdio 3. Ask your agent to do something in the browser. SAFETY • The local server listens on loopback only, so nothing on your network or the internet can reach it. • Requests are origin-checked and require a bearer token, so a website cannot drive your browser. • Chrome shows its debugging banner while an agent is connected, so you can always tell when it is active. • Page content goes only to the local server you started. Where it goes after that is decided by the AI application you choose, under its own terms. REQUIREMENTS Chrome 114 or newer, and Node.js 24 or newer for the MCP server. The extension is free and the server is open source. Privacy policy: https://cultivx.com.au/browser-control/privacy Homepage: https://cultivx.com.au/browser-control