MCP Browser Agent
A browser automation agent using the Model Context Protocol (MCP) to enable browser interactions.
MCP Browser Agent
Built at AGI House MCP Hackathon
Overview
This project is a browser automation agent that uses the Model Context Protocol (MCP) to enable browser interactions. It provides a seamless integration between Claude and browser automation capabilities through our MCP server.
Thank you to Browser-Use for their browser agent capabilities that help power our MCP server!
System Requirements
- macOS (darwin 24.2.0)
- Python 3.12 or higher
uvpackage manager- Google Chrome browser (Ensure your browser is closed before running task(s).)
Installation
Installing via Smithery
To install Browser Automation Agent for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @ashley-ha/mcp-manus --client claude
Manual Installation
- Clone the repository:
git clone <repository-url>
cd mcp
- Set up the Python environment using
uv:
uv venv
source .venv/bin/activate
uv sync
Configuration
Claude Desktop Configuration
Create or modify your Claude Desktop configuration file:
{
"mcpServers": {
"browser-use": {
"command": "uv",
"args": [
"--directory",
"/ABSOLUTE/PATH/TO/mcp",
"run",
"browser-use.py"
]
}
}
}
Replace /ABSOLUTE/PATH/TO/browser-use with the absolute path to your project directory.
Browser Configuration
The agent is configured to use Google Chrome with the following default settings:
- Non-headless mode for development
- Window size: 1280x1100
- Disabled security features for testing
- Recording path: ./tmp/recordings
Features
- Browser automation through MCP tools
- State management and planning capabilities
- Interactive element detection and manipulation
- Configurable browser contexts
- Logging and debugging support
Usage
The agent provides two main tools:
get_planner_state: Retrieves the current browser state and planning contextexecute_actions: Executes planned actions in the browser
Development
Logging
The project uses Python's built-in logging with the following configuration:
- All logs are directed to stderr
- Custom formatting:
%(levelname)-8s [%(name)s] %(message)s - Root logger level: INFO
- Third-party loggers level: WARNING
Project Structure
browser-use.py: Main entry point and server implementationtmp/recordings: Directory for browser session recordings- Dependencies managed through
uv
Contributing
This project was built during the AGI House MCP Hackathon. Contributions are welcome!
License
This project is licensed under the MIT License - see the LICENSE file for details.
Copyright (c) 2025 Jaeyun Ha, Ashley Ha
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Related Servers
MCP Deep Web Research Server
An advanced web research server with intelligent search queuing, enhanced content extraction, and deep research capabilities.
Career Site Jobs
A MCP server to retrieve up-to-date jobs from company career sites.
Notte
Leverage Notte Web AI agents & cloud browser sessions for scalable browser automation & scraping workflows
MCP-Puppeteer-Linux
Automate web browsers on Linux using Puppeteer. Enables LLMs to interact with web pages, take screenshots, and execute JavaScript.
Website to Markdown MCP Server
Fetches and converts website content to Markdown with AI-powered cleanup, OpenAPI support, and stealth browsing.
Web Search
Performs web searches and extracts full page content from search results.
Configurable Puppeteer MCP Server
A configurable MCP server for browser automation using Puppeteer.
LinkedIn
Scrape LinkedIn profiles, companies, and jobs using direct URLs. Features Claude AI integration and secure credential storage.
MCP Webscan Server
Fetch, analyze, and extract information from web pages.
Kakuyomu MCP Server
An MCP server for the Kakuyomu novel posting site, enabling users to search for works, retrieve episode lists, and read content.