agent-browser-mcp
An MCP server that provides browser automation capabilities, enabling LLMs to control a web browser for navigation, interaction, and data extraction.
README
Agent Browser MCP
A Model Context Protocol (MCP) server that provides browser automation capabilities through Vercel's agent-browser. This enables LLMs to interact with web pages using a fast Rust CLI with Node.js fallback.
Quick Start
# 1. Install agent-browser CLI
npm install -g agent-browser && agent-browser install
# 2. Add to Claude Desktop (or your MCP client)
npx agent-browser-mcp
Then use tools like browser_navigate, browser_click, browser_snapshot to control the browser from your AI agent.
Features
- AI-Optimized Browser Control - Semantic element locators using accessibility properties, text matching, and data attributes
- Session Isolation - Multiple isolated browser sessions with separate cookies, storage, and navigation history
- Comprehensive Automation - Navigation, form filling, clicking, scrolling, keyboard input, and more
- Data Extraction - Get text, HTML, attributes, accessibility snapshots, screenshots, and PDFs
- Cookie Management - Full control over browser cookies and storage
- JavaScript Execution - Run arbitrary scripts in the browser context
- Network Inspection - Monitor console messages and network requests
Use Cases
- Web Scraping for AI - Extract structured data from websites for RAG, research, or analysis
- Automated Testing - AI-powered end-to-end testing with natural language assertions
- Form Automation - Fill forms, submit data, handle multi-step workflows
- Screenshot Analysis - Capture pages for visual AI analysis or documentation
- Session Automation - Login once, persist cookies, automate authenticated workflows
- Data Entry - Bulk data input across web applications
- Monitoring - Track changes on web pages, detect updates, gather metrics
Why Agent Browser?
Agent Browser was built from the ground up for AI agents, not adapted from human-centric automation tools. This fundamental difference delivers significant advantages:
Agent-Native Design
| Capability | Agent Browser | Traditional Tools |
|---|---|---|
| Semantic Element Targeting | Native accessibility-based locators | CSS selectors / XPath |
| Token Efficiency | Structured data, minimal output | Verbose HTML/screenshots |
| Response Format | AI-optimized, parseable | Human-readable |
| Error Messages | Actionable, context-aware | Generic stack traces |
Performance
- Rust-powered CLI - Fast execution with minimal overhead
- Lightweight snapshots - Accessibility tree instead of full DOM
- Streaming output - Real-time feedback without buffering
Installation
npm install agent-browser-mcp
Or run directly with npx:
npx agent-browser-mcp
Prerequisites
- Node.js 18 or newer
- agent-browser CLI installed:
# Install agent-browser globally
npm install -g agent-browser
# Download Chromium browser
agent-browser install
# On Linux, install system dependencies if needed:
# agent-browser install --with-deps
⚠️ Windows Note: agent-browser currently has known issues on Windows with native shells (PowerShell/CMD). For Windows users, we recommend using WSL (Windows Subsystem for Linux) until the upstream issue is resolved.
Configuration
VS Code
Add to your VS Code settings (JSON):
{
"mcp": {
"servers": {
"agent-browser": {
"command": "npx",
"args": ["agent-browser-mcp"]
}
}
}
}
Claude Desktop
Add to your claude_desktop_config.json:
<details> <summary>macOS: ~/Library/Application Support/Claude/claude_desktop_config.json</summary>
{
"mcpServers": {
"agent-browser": {
"command": "npx",
"args": ["agent-browser-mcp"]
}
}
}
</details>
<details> <summary>Windows: %APPDATA%\Claude\claude_desktop_config.json</summary>
{
"mcpServers": {
"agent-browser": {
"command": "npx",
"args": ["agent-browser-mcp"]
}
}
}
</details>
Cursor
Add to Cursor MCP settings:
{
"mcpServers": {
"agent-browser": {
"command": "npx",
"args": ["agent-browser-mcp"]
}
}
}
Claude Code
claude mcp add agent-browser -- npx agent-browser-mcp
Custom agent-browser Path
If agent-browser is not in your PATH, specify its location:
{
"mcpServers": {
"agent-browser": {
"command": "npx",
"args": ["agent-browser-mcp"],
"env": {
"AGENT_BROWSER_PATH": "/path/to/agent-browser"
}
}
}
}
Available Tools
Navigation
| Tool | Description |
|---|---|
browser_navigate |
Navigate to a URL |
browser_go_back |
Navigate back in browser history |
browser_go_forward |
Navigate forward in browser history |
browser_reload |
Reload the current page |
Interaction
| Tool | Description |
|---|---|
browser_click |
Click on an element using selector or accessibility locator |
browser_fill |
Fill a text input field with a value |
browser_type |
Type text character by character (triggers key events) |
browser_hover |
Hover over an element |
browser_scroll |
Scroll the page or a specific element |
browser_select |
Select an option from a dropdown |
browser_check |
Check a checkbox or radio button |
browser_uncheck |
Uncheck a checkbox |
browser_press |
Press a keyboard key (Enter, Escape, Tab, etc.) |
Data Extraction
| Tool | Description |
|---|---|
browser_get_text |
Get text content from an element or the entire page |
browser_get_html |
Get HTML content (inner or outer) |
browser_get_attribute |
Get an attribute value from an element |
browser_get_url |
Get the current page URL |
browser_get_title |
Get the current page title |
browser_snapshot |
Get accessibility tree snapshot for AI-friendly element references |
Element State
| Tool | Description |
|---|---|
browser_is_visible |
Check if an element is visible |
browser_is_enabled |
Check if an element is enabled |
browser_is_checked |
Check if a checkbox or radio button is checked |
Screenshots & PDF
| Tool | Description |
|---|---|
browser_screenshot |
Take a screenshot of the page or a specific element |
browser_pdf |
Generate a PDF of the current page |
Session Management
| Tool | Description |
|---|---|
browser_new_session |
Create a new isolated browser session with optional viewport |
browser_close_session |
Close a browser session |
Wait Operations
| Tool | Description |
|---|---|
browser_wait_for_selector |
Wait for an element to appear (attached, detached, visible, hidden) |
browser_wait_for_navigation |
Wait for navigation to complete |
Cookies & Storage
| Tool | Description |
|---|---|
browser_get_cookies |
Get cookies, optionally filtered by URLs |
browser_set_cookies |
Set cookies with full options (domain, path, expiry, etc.) |
browser_clear_cookies |
Clear all cookies |
JavaScript & Debugging
| Tool | Description |
|---|---|
browser_evaluate |
Execute JavaScript code in the browser context |
browser_get_console |
Get console messages from the browser |
browser_get_network |
Get network requests made by the browser |
Selector Syntax
agent-browser supports semantic locators that are AI-friendly:
# By role and name
button:has-text("Submit")
[role="button"][name="Login"]
# By text content
text=Click here
:has-text("Welcome")
# By accessibility attributes
[aria-label="Search"]
[placeholder="Enter email"]
# By test IDs
[data-testid="submit-button"]
# Standard CSS selectors
#email
.form-input
form > input[type="text"]
Session Management
Create isolated browser sessions for parallel automation:
// Create a new session with custom viewport
const session = await client.callTool({
name: "browser_new_session",
arguments: {
viewport: { width: 1920, height: 1080 }
}
});
// Use session ID for subsequent operations
const result = await client.callTool({
name: "browser_navigate",
arguments: {
url: "https://example.com",
sessionId: "session-id-here"
}
});
// Close when done
await client.callTool({
name: "browser_close_session",
arguments: { sessionId: "session-id-here" }
});
Programmatic Usage
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
const transport = new StdioClientTransport({
command: "npx",
args: ["agent-browser-mcp"],
});
const client = new Client({
name: "my-browser-client",
version: "1.0.0",
});
await client.connect(transport);
// Navigate to a page
await client.callTool({
name: "browser_navigate",
arguments: { url: "https://example.com" }
});
// Get page content
const text = await client.callTool({
name: "browser_get_text",
arguments: {}
});
// Take a screenshot
await client.callTool({
name: "browser_screenshot",
arguments: {
path: "/tmp/screenshot.png",
fullPage: true
}
});
Environment Variables
| Variable | Description | Default |
|---|---|---|
AGENT_BROWSER_PATH |
Path to agent-browser executable | agent-browser |
Development
# Clone the repository
git clone https://github.com/minhlucvan/agent-browser-mcp.git
cd agent-browser-mcp
# Install dependencies
npm install
# Build
npm run build
# Run tests
npm test
# Watch mode
npm run dev
# Start server
npm start
Related Projects
- agent-browser - The underlying browser automation CLI by Vercel
- Playwright MCP - Microsoft's Playwright-based MCP server
- MCP Chrome - Chrome extension-based MCP server
- Model Context Protocol - The MCP specification
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.