MCP Chrome

Overview

MCP Chrome turns your Chrome browser into an MCP server. Unlike Browser Use or Playwright that launch separate browsers, MCP Chrome works inside your existing session — with all your logged-in accounts, cookies, and extensions.

How It Works

1. Install extension — Add from Chrome Web Store
2. Configure MCP client — Point to the extension’s endpoint
3. Use your browser — Agent reads pages, clicks, manages tabs

Use Cases

• Context-aware browsing — AI reads pages you’re viewing
• Tab management — Agent organizes open tabs
• Form filling — AI fills forms in your logged-in session
• Content extraction — Extract data from authenticated pages

Getting Started

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

Example

User: "Summarize all my open tabs"
Agent: "12 tabs: 3 GitHub PRs (2 need review), 2 Stack Overflow,
1 Jira board, 4 docs pages, 2 misc.
Suggestion: 2 SO questions are about the same topic."

Alternatives

• Browser Use — Separate browser automation
• Chrome DevTools MCP — DevTools protocol access
• Stagehand — AI browser automation