CodingBaby Browser MCP server
I. Introduction to CodingBaby Browser-Use MCP Principles
1.1 Overview
CodingBaby Browser-Use MCP (Model Context Protocol) is a tool that bridges an AI assistant with the Chrome browser, enabling the AI assistant to directly control Chrome for web browsing, interaction, and data collection. It communicates with a Chrome extension via the WebSocket protocol to achieve precise browser control.
1.2 Core Principles
The operation of CodingBaby Browser-Use MCP is based on several key components:
- MCP Server: A Node.js-based server providing standardized MCP interfaces, allowing the AI assistant to invoke browser control functions.
- WebSocket Communication: Uses the WebSocket protocol to establish a bi-directional real-time communication channel between the MCP server and the Chrome extension.
- Chrome Extension: An extension installed in the Chrome browser that receives commands from the MCP server and executes browser operations.
- Command Translation: Converts high-level commands from the AI assistant into low-level browser operations understandable by the Chrome extension.
The overall flow is as follows:
AI Assistant → MCP Server → WebSocket → Chrome Extension → Browser Operation → Web Page
1.3 Advantages
Compared to traditional browser automation tools like Puppeteer or Playwright, CodingBaby Browser-Use MCP offers several advantages:
- Higher Privileges: Running as a Chrome extension provides higher browser privileges.
- Anti-Bot Evasion: Uses a real Chrome browser and user session, effectively bypassing many website anti-bot mechanisms.
- Closer to Real User Behavior: Operations mimic real user interactions, reducing the likelihood of detection.
- Session Persistence: Can maintain login states and cookies, facilitating access to authenticated pages.
- Lightweight Deployment: Only requires installing the extension in the browser, eliminating the need for separate browser instances or WebDrivers.
II. CodingBaby Browser-Use MCP Functionality
2.1 Main Functions
CodingBaby Browser-Use MCP provides the following core functionalities:
- Web Navigation (
navigate): Open URLs, go back/forward, etc. - Element Clicking (
click): Click elements on the page via coordinates. - Text Input (
type): Input text into forms, input fields, etc. - Key Pressing (
press_key): Simulate keyboard actions. - Page Scrolling (
scroll): Scroll page content up, down, left, or right. - Screenshot Capture (
snapshot,take_screenshot): Get screenshots of the current page. - HTML Saving (
save_html): Save the current page's HTML content. - Browser Closing (
close): Close the current browser session/tab.
2.2 Technical Implementation
The core of CodingBaby Browser-Use MCP's implementation lies in the WebSocket communication mechanism:
- Port Listening: The server starts a WebSocket server on port 9876, awaiting connection from the Chrome extension.
- Message Format: Uses JSON format for communication, including command type, parameters, and request ID.
- Asynchronous Responses: Each request has a unique ID to ensure correct matching of requests and responses.
- Error Handling: Includes timeout and error handling mechanisms for communication stability.
- Session Management: Capable of managing multiple browser tab sessions (depending on extension implementation).
III. Installation and Configuration
3.1 Prerequisites
- Node.js 16+
- Chrome Browser
- The companion Chrome Extension (Chrome MCP Extension) installed and enabled.
3.2 Setup Steps
-
Set up CodingBaby Browser-Use MCP (if running from source):
# Install dependencies npm install # Start the server node chrome-server.jsOr install the package:
npm install @sydneyassistent/codingbaby-browser-mcp -
Register MCP tools in your environment (e.g., Cursor):
Edit
.cursor/mcp.jsonto add/update the server configuration:{ "mcpServers": { "chrome-server": { "command": "npx", "args": ["@sydneyassistent/codingbaby-browser-mcp"] } // ... other servers ... } }
Remember to restart your MCP environment after changing configurations.
Server Config
{
"mcpServers": {
"codingbaby-browser-mcp": {
"command": "npx",
"args": [
"-y",
"@sydneyassistent/codingbaby-browser-mcp"
]
}
}
}Recommend Servers
View AllWrite notes to Flomo