Chrome MCP Server

Chrome browser sessions and navigation tools for AI-driven automation tasks.

Data last scanned today · Reviewed today

44 4 dangerous tools28 tools★ 12,213TypeScriptMIT

Overview

Chrome MCP Server acts as a bridge between your active Chrome browser and AI models. By functioning as a browser extension, it allows AI to interact directly with your existing browser instance, maintaining your current session, login states, and configurations for a seamless transition between human and AI interaction. The server provides over 20 distinct tools covering operations like network monitoring, bookmark management, history access, and visual analysis through screenshots. It features cross-tab context capabilities and uses local WebAssembly SIMD-accelerated vector databases to enable semantic search across your browsing history and open tabs.

Our verdict

This server carries a high-risk score of 44/100, primarily due to the lack of external security reviews for its authentication and read-only implementations. While it offers deep integration with personal browser data, users should be cautious given that it is a community-maintained project and has not received updates in over 200 days. It is suitable for experimental use in non-sensitive environments, but not recommended for workflows involving private or highly sensitive browser data.

Operates entirely locally for improved privacy
Uses existing browser login states and configurations
Includes over 20 specialized browser automation tools
Features SIMD-accelerated local vector database for search

High risk score due to unreviewed OAuth and authentication
Lacks recent updates with last commit over 200 days ago
Community-maintained rather than official browser vendor support
Read-only capabilities have not been editorially verified

Setup

Claude Desktop

{
  "mcpServers": {
    "mcp-chrome": {
      "args": [
        "-y",
        "mcp-chrome-bridge-monorepo"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
get_windows_and_tabs	List all currently open browser windows and tabs.	low
chrome_navigate	Navigate to a URL with optional viewport control.	medium
chrome_switch_tab	Switch the current active tab.	medium
chrome_close_tabs	Close specific tabs or windows.	high
chrome_go_back_or_forward	Browser navigation control.	medium
chrome_inject_script	Inject content scripts into web pages.	high
chrome_send_command_to_inject_script	Send commands to injected content scripts.	medium
chrome_screenshot	Advanced screenshot capture with element targeting, full-page support, and custom dimensions.	low
chrome_network_capture_start/stop	webRequest API network capture.	low
chrome_network_debugger_start/stop	Debugger API with response bodies.	low

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›What are the hardware and software prerequisites for this server?

You need Node.js version 20.0.0 or higher, a package manager like npm or pnpm, and a Chrome or Chromium-based browser.

›How does this differ from Playwright-based MCP servers?

Unlike Playwright-based servers, this extension-based approach does not launch an independent browser process, automatically uses your existing login sessions and browser configurations, and provides access to native Chrome APIs.

›What should I do if the automatic registration fails during installation?

If automatic registration fails, you can manually register the bridge by running the command 'mcp-chrome-bridge register' in your terminal.

›How is the browser content indexed for semantic search?

The server features a built-in vector database with custom WebAssembly SIMD optimization, which enables intelligent tab content discovery and text extraction.

›Is this project considered stable for production use?

The project is currently in its early stages and is under intensive development, meaning future updates will focus on stability and new feature enhancements.