Browser Agent

A robust browser automation agent designed to interact with web pages programmatically, exposed via a FastAPI web interface. This project leverages Playwright for browser control and mcp-agent for core agent functionalities, providing a powerful and flexible solution for web scraping, automated testing, and other browser-based tasks.

Demo

Watch the Browser Agent in action:

browser.mp4

Features

• MCP Support: MCP from playwright
• Asynchronous Operations: Built with asyncio and uvloop for high performance and concurrency.
• Configurable: Easily customize agent behavior through mcp_agent.config.yaml.
• Modular Design: Integrates with mcp-agent for extensible agent capabilities.

Installation

To set up the Browser Agent, follow these steps:

1. Clone the repository:

   git clone https://github.com/priyank766/Browser_Agent_MCP.git
   cd browser-agent

2. Install dependencies:

   uv add -r requirements.txt

   OR

   pip install -r requirements.txt

3. Check Node and NPM:

   1. node --version
   2. npm --version

4. Paste API Keys

   Paste Google API Key in mcp_agent.secrets.yaml Choose Model IN mcp_agents.config.yaml

Usage

To start the Browser Agent, run the following command:

streamlit run browser_agents.py

This will start the Web application, typically accessible at http://127.0.0.1:8000.

You can then change the agent provider from mcp_agent.config.yaml Refer to the browser_agent.py file

Configuration

The agent's behavior can be configured using the mcp_agent.config.yaml file. This file allows you to set various parameters for the browser, agent behavior, and other settings.

Contributing

Contributions are welcome! Please feel free to submit pull requests or open issues for bugs and feature requests.

License

This project is licensed under the MIT License. See the LICENSE file for details.