Browser MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Browser MCP Server Navigate to https://example.com and analyze the page structure."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Browser MCP Server
A Model Context Protocol (MCP) server that enables AI agents to understand web page structure and content without screenshots. Built with Playwright for reliable browser automation.
šÆ Purpose
Solves the problem where AI agents give "wrong and random suggestions" about web pages by providing them with structured page data instead of requiring visual screenshots.
Related MCP server: Playwright MCP
⨠Features
š Page Structure Analysis - Understand layout, forms, and interactive elements
š Smart Content Extraction - Extract text using CSS selectors
šÆ Element Discovery - Find and analyze specific page elements with accessibility info
š Fast & Local - Uses Playwright's accessibility tree (no screenshots needed)
š¤ AI-Optimized - Designed specifically for AI agent integration
š¦ Installation
npm install @your-org/browser-mcp-serverš Usage
Claude Desktop Integration
Add to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"browser": {
"command": "npx",
"args": ["@your-org/browser-mcp-server"]
}
}
}VS Code + Copilot Chat
Configure in your VS Code MCP settings to use with Copilot Chat.
Windsurf Integration
Add to your Windsurf MCP configuration.
Local Development
# Link for local development
npm link @your-org/browser-mcp-server
# Use in your project
const browserMCP = require('@your-org/browser-mcp-server');š ļø Available Tools
browser_navigate
Navigate to a URL and get page information.
{
"url": "https://example.com"
}browser_analyze_page
Get comprehensive page structure analysis.
{
"includeContent": true,
"includeForms": true,
"includeLinks": true,
"maxDepth": 2
}browser_extract_text
Extract text content from page or specific elements.
{
"selector": "h1, h2, p",
"includeHidden": false
}browser_get_elements
Find elements with detailed attributes and accessibility info.
{
"selector": "button, input, a[href]",
"includeAttributes": true,
"maxResults": 20
}š Example AI Conversations
Understanding Page Flow
Human: Go to https://github.com/login and help me understand the login process
AI: I'll navigate to the GitHub login page and analyze its structure.
[Uses browser_navigate and browser_analyze_page]
The login page has:
- 2 input fields: username/email and password (both required)
- 1 "Sign in" submit button
- Links for "Forgot password" and "Create account"
- Form submits to /session with POST method
- Additional 2FA field may appear after initial submission
The typical flow is: enter credentials ā click Sign in ā (optional 2FA) ā redirect to dashboard.Content Extraction
Human: Extract all the main article titles from https://news.ycombinator.com
AI: [Uses browser_navigate and browser_extract_text with CSS selectors]
Found 30 article titles including:
1. "Show HN: My new startup idea"
2. "Why distributed systems are hard"
3. "The future of web development"
[etc...]š§ Development
# Clone and setup
git clone <your-repo>
cd browser-mcp-server
npm install
# Install browser
npx playwright install chromium
# Build
npm run build
# Test locally
npm startšÆ Perfect For
AI Agents that need to understand web page workflows
Automated Analysis of website structure
Content Extraction without visual parsing
Form Discovery for automation planning
Accessibility Analysis of web pages
šļø Architecture
āāā src/
ā āāā index.ts # MCP server entry point
ā āāā server.ts # MCP server implementation
ā āāā browser/ # Browser management
ā ā āāā manager.ts # Playwright browser lifecycle
ā āāā tools/ # MCP tools
ā ā āāā navigate.ts # Page navigation
ā ā āāā analyze.ts # Page structure analysis
ā ā āāā extract.ts # Content extraction
ā ā āāā elements.ts # Element discovery
ā āāā types.ts # TypeScript definitionsš Requirements
Node.js 18+
Chromium (auto-installed via Playwright)
MCP-compatible AI agent (Claude Desktop, VS Code, Windsurf, etc.)
š License
MIT
Built for AI agents to understand the web, not just see it. š¤š
Available Tools
4 toolsbrowser_analyze_pageB
Analyze the current page structure, forms, and interactive elements to help understand page flow and functionality
Inputs: maxDepth, includeForms, includeLinks, includeContent.
browser_extract_textA
Extract text content from the current page or specific elements using CSS selectors
Inputs: selector, includeHidden.
browser_get_elementsA
Get detailed information about elements matching a CSS selector, useful for understanding page structure and interactive elements
Inputs: selector (required), maxResults, includeAttributes.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelogā¦
TDQS
Scored across 4 tools
Each tool has a distinct purpose: navigation, high-level page analysis, text extraction, and element inspection. There is slight overlap between analyze_page and get_elements, but the descriptions clarify that one provides an overview while the other targets specific selectors.
All tools follow a consistent 'browser_verb_noun' pattern (navigate, analyze_page, extract_text, get_elements). This makes the set predictable and easy to understand.
With only 4 tools, the server is well-scoped for its apparent read-only browser analysis purpose. Each tool is necessary and there is no redundancy.
The tool set covers the core workflow of navigating to a page and extracting both text and structural information. It lacks interaction capabilities like clicking or typing, but for a read-only analysis server this is a minor gap rather than a fatal omission.
Maintenance
Related MCP Connectors
Playwright browser automation ā navigate, click, fill forms, extract, screenshot ā 17 agent tools.
Hosted browser for AI agents: screenshots, post-JS DOM, console, WCAG. No install, no API key.
Give agents eyes on any web page: structured context, and changes explained in plain language.
Cloud browser, web extraction, screenshots, and remote CDP tools for AI agents.
Related MCP Servers
- -licenseNot gradedqualityNot gradedmaintenanceEnables browser automation and web interaction through structured accessibility snapshots using Playwright. Provides fast, deterministic web page interaction without requiring screenshots or vision models.10,759,741 npm-
- AlicenseNot gradedqualityDmaintenanceEnables LLMs to perform browser automation and web page interactions using Playwright's accessibility tree instead of screenshots. Provides fast, deterministic web automation through structured data without requiring vision models.10,759,741 npmApache 2.0
- AlicenseNot gradedqualityDmaintenanceEnables browser automation and web page interaction through Playwright's accessibility tree, allowing LLMs to navigate, fill forms, click elements, and extract content without requiring vision models or screenshots.10,759,741 npmApache 2.0
- AlicenseNot gradedqualityDmaintenanceEnables web browser automation and inspection using structured data instead of screenshots, allowing AI agents to interact with web pages programmatically through the Playwright framework.ISC