Auto-Browser
🐍 🏠 🐧 🍎 🪟 - Open-source MCP-native browser agent with human takeover via noVNC, re
What is Auto-Browser?
Auto-Browser is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to 🐍 🏠 🐧 🍎 🪟 - open-source mcp-native browser agent with human takeover via novnc, re
🐍 🏠 🐧 🍎 🪟 - Open-source MCP-native browser agent with human takeover via noVNC, re
This server falls under the Browser Automation category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- MCP protocol support
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx auto-browserConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Auto-Browser
Auto Browser is an MCP-native browser control plane that gives AI agents a real Playwright-backed browser with human-in-the-loop takeover capabilities via noVNC. It exposes browser sessions as MCP tools — allowing Claude and other MCP clients to create sessions, navigate pages, take screenshots, inspect the DOM, and save named authentication profiles for reuse. Teams use it for operator-assisted QA, internal dashboard automation, and login-once-reuse-later workflows where brittle sites may require a human to step in without losing the active session.
Prerequisites
- Docker and Docker Compose installed (the full stack runs in containers)
- Python 3.11 or later if using the PyPI client SDK or the stdio bridge
- Git to clone the repository (the Docker Compose setup requires the repo)
- An MCP client such as Claude Desktop or Cursor
- Port 8000 (API and MCP), 6080 (noVNC) available on localhost
Clone the repository and start the stack
Clone the auto-browser repository and bring up the full stack with Docker Compose. All ports bind to 127.0.0.1 by default for local-first security.
git clone https://github.com/LvcidPsyche/auto-browser.git
cd auto-browser
docker compose up --buildVerify the services are running
Check that the API, dashboard, and noVNC are accessible. Run make doctor from a normal terminal for a full diagnostic.
# API docs: http://127.0.0.1:8000/docs
# Dashboard: http://127.0.0.1:8000/dashboard
# noVNC takeover: http://127.0.0.1:6080/vnc.html?autoconnect=true&resize=scale
make doctorInstall the stdio MCP bridge
Install the auto-browser-mcp PyPI package to get the stdio bridge that MCP clients like Claude Desktop can connect to without HTTP.
pip install auto-browser-client
# The stdio bridge runs as:
uvx auto-browser-mcpAdd Auto Browser to your MCP client configuration
Add the stdio bridge to your claude_desktop_config.json. Make sure the Docker stack is running before starting Claude Desktop.
{
"mcpServers": {
"auto-browser": {
"command": "uvx",
"args": ["auto-browser-mcp"]
}
}
}Create your first browser session
Use Claude or the REST API to create a browser session. You can also log in manually via noVNC and save that authenticated state as a named profile.
curl -s http://127.0.0.1:8000/sessions \
-X POST \
-H 'content-type: application/json' \
-d '{"name":"demo","start_url":"https://example.com"}' | jqAuto-Browser Examples
Client configuration
Claude Desktop configuration using the uvx stdio bridge. The Docker Compose stack must be running for the bridge to reach the browser controller.
{
"mcpServers": {
"auto-browser": {
"command": "uvx",
"args": ["auto-browser-mcp"],
"env": {
"MCP_TOOL_PROFILE": "curated"
}
}
}
}Prompts to try
Tell Claude to control the browser through Auto Browser's MCP tools.
- "Create a browser session, navigate to https://news.ycombinator.com, and take a screenshot of the top 5 stories"
- "Log in to my admin dashboard, save the session as a named auth profile called 'admin', and close the session"
- "Open a new session using the 'admin' auth profile and navigate to the user management page"
- "Take a screenshot of the current page and describe what you see in the main content area"Troubleshooting Auto-Browser
Docker Compose fails to start or ports 8000 or 6080 are already in use
Check for conflicting services with 'lsof -i :8000' and 'lsof -i :6080'. Kill any conflicting process. You can override ports in a .env file by setting API_PORT and VNC_PORT before running docker compose up.
Browser sessions are created but screenshots show a blank or loading page
Some sites take time to load. Add a wait after navigation before taking a screenshot. Use the observe tool to get the DOM state and check if content is present before screenshotting.
noVNC takeover shows 'connection refused' when trying to manually control the browser
Verify the stack is fully up by checking 'docker compose ps' — all containers should show 'running'. The noVNC service may take 5-10 seconds longer to start than the API. Refresh the noVNC page after a moment.
Frequently Asked Questions about Auto-Browser
What is Auto-Browser?
Auto-Browser is a Model Context Protocol (MCP) server that 🐍 🏠 🐧 🍎 🪟 - open-source mcp-native browser agent with human takeover via novnc, re It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Auto-Browser?
Follow the installation instructions on the Auto-Browser GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Auto-Browser?
Auto-Browser works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Auto-Browser free to use?
Yes, Auto-Browser is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.
Auto-Browser Alternatives — Similar Browser Automation Servers
Looking for alternatives to Auto-Browser? Here are other popular browser automation servers you can use with Claude, Cursor, and VS Code.
Chrome DevTools MCP
★ 40.6kAI-powered Chrome automation server with natural language element detection. Control Chrome browser through MCP protocol for testing, debugging, and performance analysis. Features 91% accuracy in element location, works with free AI models, and suppo
UI TARS Desktop
★ 34.9k📇 🏠 - Browser automation capabilities using Puppeteer, both support local and remote browser connection.
Playwright
★ 32.8kA production-ready browser automation server that enables AI assistants to interact with web pages using tools for navigation, element interaction, and data extraction. It features a built-in Inspector UI and robust crash recovery for reliable automa
Page Agent
★ 18.0kJavaScript in-page GUI agent. Control web interfaces with natural language.
Chrome
★ 11.7kAn extension-based MCP server that enables AI assistants to control your browser, leveraging existing sessions and login states for automation and content analysis. It provides over 20 tools for semantic tab search, interactive element manipulation,
LAMDA
★ 7.8kThe most powerful Android RPA agent framework, next generation mobile automation.
Browse More Browser Automation MCP Servers
Explore all browser automation servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Auto-Browser in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Auto-Browser?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.