Chrome MCP Server turns your existing Chrome browser into an intelligent automation partner. It exposes browser capabilities to AI assistants, enabling browser control, content analysis, and semantic search while preserving your current login state and extensions for a seamless, private, local workflow.

How to use

You connect a compatible MCP client to the server running in your browser, then issue browser automation and analysis requests through that client. The server integrates with your current Chrome environment, so your login state, extensions, and preferences remain active. Use the streamable HTTP option for the most straightforward setup and fastest responses, or choose a local stdio approach if you prefer to run the server directly on your machine.

Typical workflows include navigating to pages, taking screenshots, inspecting network activity, extracting content for analysis, and performing targeted actions in response to AI prompts. The system provides semantic search across open tabs, AI-assisted content extraction, and a suite of browser tools to manage tabs, history, bookmarks, and more.

How to install

Prerequisites: You need Node.js 18.19.0 or newer and a Chrome/Chromium browser.

1. Download the latest Chrome extension package from the official release channel.

2. Install the extension bridge globally on your system.

npm install -g mcp-chrome-bridge

pnpm install -g mcp-chrome-bridge

If postinstall scripts are blocked by your pnpm version, enable them or use manual registration

pnpm config set enable-pre-post-scripts true pnpm install -g mcp-chrome-bridge pnpm install -g mcp-chrome-bridge mcp-chrome-bridge register

## Configuration and connectivity

You can connect to the MCP server via two methods. The preferred method uses a Streamable HTTP connection to run the MCP server alongside your client. Alternatively, you can run a local stdio-based server that executes the MCP bridge directly on your machine.

Streamable HTTP connection example (recommended): copy this into your MCP client configuration to connect to the browser MCP server.

{ "mcpServers": { "chrome-mcp-server": { "type": "streamableHttp", "url": "http://127.0.0.1:12306/mcp", "args": [] } } }

## Local stdio configuration

If you prefer running the server locally with stdio, use the provided command and path to the mcp-server script. This runs a self-contained process that your MCP client can invoke directly.

{ "mcpServers": { "chrome-mcp-stdio": { "type": "stdio", "name": "chrome_mcp_stdio", "command": "npx", "args": [ "node", "/Users/xxx/Library/pnpm/global/5/node_modules/mcp-chrome-bridge/dist/mcp/mcp-server-stdio.js" ] } } }

## Notes and troubleshooting

This server runs locally and keeps your data private. If you encounter startup issues, ensure Node.js is correctly installed and the extension bridge is registered. For the best performance, use the Streamable HTTP method when your client supports it, and verify that Chrome is running with the extension enabled.

## Troubleshooting

If you see connection failures, confirm the MCP bridge is loaded and listening on the expected port. Check that Chrome’s extension is loaded and the browser session is active. Review any firewall settings that might block localhost connections.

## Security and privacy

All operations are performed locally in your environment. Your browser data, login state, and extensions remain in your control, with no data sent to external servers unless you explicitly configure a remote MCP endpoint.

## Available tools

### chrome\_navigate

Navigate to URLs and control the viewport within browser tabs.

### chrome\_close\_tabs

Close specific tabs or windows to manage your browsing sessions.

### chrome\_go\_back\_or\_forward

Navigate backward or forward in the browser history.

### chrome\_screenshot

Capture screenshots with advanced options, including full-page and element-specific captures.

### chrome\_network\_capture\_start/stop

Start and stop network capture via the webRequest API to monitor traffic.

### chrome\_network\_debugger\_start/stop

Attach to the debugger API to inspect responses and request details.

### chrome\_network\_request

Send custom HTTP requests from the browser context.

### search\_tabs\_content

Perform AI-powered semantic search across content in open tabs.

### chrome\_get\_web\_content

Extract HTML or text content from the current page.

### chrome\_get\_interactive\_elements

Identify clickable elements for automated interactions.

### chrome\_console

Capture and retrieve console output from browser tabs.

### chrome\_click\_element

Click elements using CSS selectors to simulate user interactions.

### chrome\_fill\_or\_select

Fill form fields or select options programmatically.

### chrome\_keyboard

Simulate keyboard input and shortcuts.

### chrome\_history

Search browser history with time-based filters.

### chrome\_bookmark\_search

Find bookmarks by keywords.

### chrome\_bookmark\_add

Add new bookmarks with optional folder paths.

### chrome\_bookmark\_delete

Delete bookmarks from the browser.