Voxta MCP Bridge Provider

A Voxta provider that enables communication with Model Context Protocol (MCP) servers, allowing Voxta to leverage external tools and resources through the MCP protocol.

Prerequisites

• .NET 8.0 SDK
• Python 3.10 or higher
• Git

Setup Instructions

Windows

1. Clone the repository:

git clone https://github.com/voxta/voxta-mcp.git
   cd voxta-mcp

2. Create Python virtual environment:

python -m venv venv
   .\venv\Scripts\activate
   pip install mcp-agent

3. Build and run the project:

dotnet build
   dotnet run

Linux

1. Clone the repository:

git clone https://github.com/voxta/voxta-mcp.git
   cd voxta-mcp

2. Create Python virtual environment:

python3 -m venv venv
   source venv/bin/activate
   pip install mcp-agent

3. Build and run the project:

dotnet build
   dotnet run

Running the Provider

1. Ensure your MCP server is running and accessible at the configured address.

2. Configure appsettings.json with your settings:

{
     "MCPBridge": {
       "PythonExePath": "venv/bin/python",  // Path to Python in virtual environment
       "MCPClientScriptPath": "mcp_client.py",  // Path to the MCP client script
       "MCPServerAddress": "localhost:50051"  // Your MCP server address
     }
   }

3. Run the provider:

dotnet run

4. The provider will:

• Start the Python MCP client process
• Connect to your MCP server
• Register with Voxta
• Begin handling action requests

5. Monitor the console output for:

• Connection status
• Action triggers
• Error messages
• MCP tool responses

Troubleshooting

Common issues and solutions:

1. "MCPClientScriptPath is not configured"

• Ensure appsettings.json exists and has the correct path to mcp_client.py

2. Python process fails to start

• Verify Python virtual environment is activated
• Check PythonExePath in configuration
• Ensure mcp-agent is installed in the virtual environment

3. Cannot connect to MCP server

• Verify the server is running
• Check MCPServerAddress configuration
• Ensure no firewall is blocking the connection

Configuration

The provider is configured through appsettings.json:

{
  "MCPBridge": {
    "PythonExePath": "venv/bin/python",  // On Windows: "venv\\Scripts\\python.exe"
    "MCPClientScriptPath": "mcp_client.py",  // Required: Path to the Python MCP client script
    "MCPServerAddress": "localhost:50051"  // Address of your MCP server
  },
  "Serilog": {
    "Using": ["Serilog.Sinks.Console"],
    "WriteTo": [
      {
        "Name": "Console",
        "Args": {
          "outputTemplate": "[{Timestamp:HH:mm:ss} {Level:u3}] {Message:l}{NewLine}{Exception}",
          "theme": "Serilog.Sinks.SystemConsole.Themes.SystemConsoleTheme::Literate, Serilog.Sinks.Console"
        }
      }
    ],
    "MinimumLevel": {
      "Default": "Information",
      "Override": {
        "Microsoft": "Warning",
        "System.Net.Http.HttpClient": "Warning",
        "Voxta": "Information",
        "Voxta.MCPBridgeProvider": "Debug"
      }
    }
  }
}

Required Configuration

• MCPBridge:MCPClientScriptPath: Path to the Python MCP client script that handles communication with the MCP server.
• MCPBridge:MCPServerAddress: Address of your MCP server (e.g., "localhost:50051" for a local server).

Optional Configuration

• MCPBridge:PythonExePath: Path to the Python executable. Defaults to "python" (or "python3" on Linux).

Connecting to MCP Servers

1. Start your MCP server (e.g., Home Assistant MCP server) and note its address (typically localhost:50051).

2. Update the MCPServerAddress in appsettings.json to match your MCP server's address.

3. The provider will automatically connect to the MCP server when started and make its tools available to Voxta.

Usage

1. The provider will automatically register available MCP tools with Voxta.

2. When Voxta triggers an action, the provider will:

• Translate the Voxta action to an MCP tool call
• Send the request to the MCP server through the Python client
• Receive the response and send it back to Voxta

Development

• The C# code handles the Voxta integration and process management
• The Python script (mcp_client.py) handles MCP protocol communication
• Both components communicate through stdin/stdout using JSON messages

License

[License information here]