Repolode
Submit
leeguooooo

leeguooooo/chrome-use

A Rust CLI and MCP server that lets AI agents drive your real, logged-in Chrome through a local native-messaging extension.

View on GitHubWebsite
Stars
231
Forks
24
Open issues
8
Last commit
12 hours ago

What is it?

chrome-use lets any AI agent, such as Claude Code, Cursor or Codex, control the Chrome browser you already use, with your existing logins, cookies and profile intact. A small extension talks to a local CLI over Chrome native messaging, so no debug port, token or remote server is involved, and no per-use permission prompt appears.

Each --session gets its own colored tab group, so several agents can work in one browser without interfering with each other or with your own tabs. The tool exposes page snapshots based on the DOM and accessibility tree, and it can also run as an MCP stdio server for hosts that cannot run shell commands. The project includes a side panel for chatting with local agent CLIs, an element inspector that sends selected page elements to a coding session, and a recorder that turns a manual flow into a replayable script.

Installation is a one-line shell or PowerShell script that fetches a prebuilt binary, and the project is licensed under Apache-2.0. It runs on macOS, Linux and Windows and originated as a fork of vercel-labs/agent-browser.

Key features

  • 01Drives your existing logged-in Chrome through a native-messaging extension without a debug port.
  • 02Gives each agent session its own colored tab group for isolated concurrent use.
  • 03Provides DOM and accessibility snapshots with element refs for acting on pages.
  • 04Runs as an MCP stdio server for clients that cannot execute shell commands.
  • 05Includes a side panel for chatting with local Codex, Claude Code, OpenCode or Pi CLIs.

Use cases

  • Letting an AI agent operate a website where you are already signed in, such as a dashboard or social account.
  • Running several coding agents in parallel against one shared browser with separate tab groups.
  • Extracting structured data from pages that block fresh automated browsers.
  • Recording a manual checkout or form flow and replaying it without a model.

Quick start

  1. 1

    Run the official install script from the README (macOS/Linux)

  2. 2

    Connect the extension relay to your Chrome

    chrome-use extension connect
  3. 3

    Install the bundled AI agent skill

    chrome-use skills install

Things to know

  • Chrome shows its 'started debugging this browser' bar while an agent tab is attached, because the extension uses chrome.debugger.
  • Setup requires installing a browser extension and registering a native-messaging host before it works.

In collections

Browser automation for AI agentsMCP servers for Claude CodeClaude Code tools, skills & pluginsOpen-source AI agent frameworks & appsMCP serversCLI & terminal toolsChrome & browser extensions

Topics

#ai-agent#browser-automation#chrome-extension#mcp#cli#rust#model-context-protocol#anti-detection#chrome#claude

FAQ

What is chrome-use?
A Rust CLI and MCP server that lets AI agents drive your real, logged-in Chrome through a local native-messaging extension. chrome-use lets any AI agent, such as Claude Code, Cursor or Codex, control the Chrome browser you already use, with your existing logins, cookies and profile intact. A small extension talks to a local CLI over Chrome native messaging, so no debug port, token or remote server is involved, and no per-use permission prompt appears.
Is chrome-use free to use? What license does it have?
chrome-use is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is chrome-use still maintained?
Actively maintained. The last commit to chrome-use was on Nov 6, 2026.
What can I use chrome-use for?
Letting an AI agent operate a website where you are already signed in, such as a dashboard or social account. Running several coding agents in parallel against one shared browser with separate tab groups. Extracting structured data from pages that block fresh automated browsers. Recording a manual checkout or form flow and replaying it without a model.
How do I install chrome-use?
Run the official install script from the README (macOS/Linux) Connect the extension relay to your Chrome: chrome-use extension connect Install the bundled AI agent skill: chrome-use skills install
Where can I get chrome-use?
The source code and installation instructions are on GitHub: https://github.com/leeguooooo/chrome-use

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects