Local Browser: AI Browser Automation

Let your AI agent use your existing Chrome tabs and logins. Read pages, fill forms and take screenshots through MCP.

Put your AI agent to work in the Chrome browser you already use. Local Browser connects your agent to your existing tabs and signed-in accounts. Ask it to read a page, fill a form, click a button or capture a screenshot, without moving your work to a separate browser or signing in again. USE THE BROWSER YOU ALREADY HAVE • Read page text and find buttons, links and form fields. • Fill inputs, choose options and click through browser tasks. • Open, switch and organize tabs. • Run custom page JavaScript supplied by your connected agent. • Take screenshots so your agent can check the result. CONNECT YOUR OWN AGENT Built for Hermes Agent, with an MCP server for other MCP clients. MCP lets your agent call browser tools through a standard connection. Local Browser uses Chrome extension APIs and does not need a remote-debugging port. YOU CONTROL THE CONNECTION Pair the extension with your own relay. See its connection state and last task in the popup. Pause new browser commands whenever you want, reconnect, or forget the connection. SETUP This extension requires the free, open-source Local Browser Python relay and MCP setup on your agent's computer. Chrome and the agent can run on the same computer, or connect privately across computers with Tailscale and WSS. Import the pairing file from setup and click Connect my agent. Enable Allow User Scripts in Chrome's extension settings for custom JavaScript. Start here: https://github.com/codejeet/local-browser YOUR DATA Requested page content and screenshots go to the relay and agent you choose, and may reach the agent's model provider or logs. Connect only to an agent you trust. Local Browser does not send browser content to a developer-run analytics or advertising service. Privacy: https://github.com/codejeet/local-browser/blob/main/docs/privacy.md