Short answers about setup, safety, profiles and parallel tabs.
WebBridge, published as agent-webbridge, is an open-source (MIT) browser-automation MCP server and Chrome extension. It lets AI agents such as Claude Code or Cursor use your own signed-in Chrome profiles, with several tabs and profiles running at the same time. Everything runs on your computer (127.0.0.1).
Not to install it. Paste the setup prompt into an AI assistant that can run terminal commands, such as Claude Code or Cursor. You need Google Chrome and Node.js 18 or newer, and you click "Add to Chrome" yourself. A plain chat window that cannot run commands or connect MCP servers cannot set it up.
It is powerful, so treat it like handing someone your browser. Since version 1.3.1 the bridge refuses requests that come from web pages or from non-local hosts, and versions before 1.3.1 had a security hole, so update. The AI acts as you: watch what it does, ask it to stop before submitting anything important, and use it only on accounts and sites you are allowed to automate.
Any app that supports MCP servers, or that can send HTTP requests to the local router. Setup instructions are written for Claude Code, Claude Desktop, Cursor and Windsurf.
Yes. Each profile gets its own local daemon, and a router sends each request to the profile you name. Inside a profile every tab is controlled separately, so tabs run at the same time. In our benchmark on one MacBook, 5 tabs finished in 3.5 seconds in parallel versus 9.3 seconds one at a time; results vary by machine and site.
Playwright and cloud browsers usually start a fresh browser, so you log in again and deal with 2FA. WebBridge uses the Chrome you already use, so existing sessions are reused. It does not make automation undetectable: sites may still limit automated behavior, and you must follow their terms.
No, there is no recorder yet. You describe the task to your AI, and community recipes (awb-recipes) capture tested steps for specific sites.
No. agent-webbridge (also written WebBridge or Agent WebBridge) is an independent open-source project at github.com/jeet-dhandha/agent-webbridge. It is not affiliated with other projects or companies that use the name WebBridge, including the Open WebBridge project.
Run npm i -g agent-webbridge && awb setup "Work" once to install the extension and daemon. Then in Claude Code run claude mcp add chrome -- npx -y agent-webbridge mcp, or add the stdio server config to your Claude Desktop, Cursor or Windsurf settings: {"mcpServers":{"chrome":{"command":"npx","args":["-y","agent-webbridge","mcp"]}}}.
The extension attaches Chrome's debugger separately to each tab it controls, keyed by tab ID, so a slow page in one tab does not hold up the others. Simpler bridges that send every command through one global current tab can only run one tab at a time. In our benchmark on one MacBook, 5 tabs finished in about 3.5 seconds in parallel versus 9.3 seconds one at a time; results vary by machine and site, and opening many tabs on one site can trigger its rate limits.
awb-recipes is a separate open-source repository of community-maintained recipes: documented steps, selector fallbacks and caveats for specific tasks on specific sites. Sites change their pages, so each recipe should carry a verified date; treat an old recipe as a starting point, not a guarantee.
It is published on npm and the Chrome Web Store, and listed in the Official MCP Registry (which currently shows version 1.3.1). Pull requests to several awesome lists are open and waiting for review. The Ecosystem page shows the current status of each.
WebBridge itself is built to run locally: the router and daemons listen on 127.0.0.1, with no account, no telemetry and no remote server. From version 1.3.1 they refuse requests that carry a web-page Origin or a non-local Host, which closes a hole in earlier versions, so update. Your AI app may still send page content to its own model provider, and the AI acts as you in your signed-in browser.