MCP Aoai Web Browsing

🚀 Implementation of MCP Server and Client Using Azure OpenAI

• A minimal server/client application implementation based on the Model Context Protocol (MCP) and Azure OpenAI.
  1. The MCP server is built using FastMCP.
  2. Playwright is a modern, open - source end - to - end testing framework from Microsoft.
  3. The tool information in the MCP response will be converted to the OpenAI function call format.
  4. A custom MCP - LLM Bridge is implemented to convert the MCP server response to the OpenAI function call format.
  5. To ensure a stable connection, the server object is directly passed to the bridge.

✨ Features

Model Context Protocol (MCP)

The Model Context Protocol (MCP) is an open protocol that enables secure and controlled interactions between AI applications and their local or remote resources.

Official Repositories

• MCP Python SDK
• Create Python Server
• MCP Servers

Community Resources

• Awesome MCP Servers
• MCP on Reddit

Related Projects

• FastMCP: A fast and Pythonic way to build MCP servers.
• Chat MCP: An MCP client.
• [MCP - LLM Bridge](https://github.com/kaiyujiang/MCP - LLM - Bridge): A bridge to connect MCP to various LLM services.

📦 Installation

Configuration Steps

1. Create a .env file: Create a .env file in the project root directory and add the following content:

AZURE_OPEN_AI_ENDPOINT=your_endpoint
AZURE_OPEN_AI_KEY=your_key
MCP_SERVER_NAME=your_server_name

2. Install dependencies: Execute the following command to install the required Python packages:

pip install fastapi playwright[aiohttp] python - dotenv

3. Start the server: Execute the following command to start the MCP server:

python mcp_server.py

💻 Usage Examples

Basic Usage

Here is a pseudo - code example of the playwright_navigate function:

@self.mcp.tool()
async def playwright_navigate(url: str, timeout=30000, wait_until="load"):
    # Implement the specific navigation logic here
    pass

📚 Documentation

Tips

• Run the command python chatgui.py to start the chat interface.
• Use Ctrl + C to exit the program.
• If you need to reload the configuration, you can restart the server.

Image Example

Width: 300