BingHanLin/CPP_MCP-Bridge-Demo

C++ MCP Bridge Demo

This project demonstrates how to create a bridge between C++ applications and AI assistants using a Python MCP server with support for both Socket and gRPC communication.

Project Structure

• cpp_app/: Core C++ application with business logic
• mcp_server/: Python MCP server implementation
• proto/: Protocol buffer definitions for gRPC

Features

• Python MCP Server: Implementing Model Context Protocol for AI assistant integration
• C++ Application: Core business logic with command processing
• Dual Communication: Both Socket and gRPC protocols supported

Quick Start

Prerequisites

• C++ Compiler: Visual Studio 2019+ (tested on Windows)
• CMake: 3.10 or higher
• Python: 3.10 or higher with UV package manager

Build & Setup

# 1. Clone and navigate to the project
git clone <repository-url>
cd cpp-mcp-bridge-demo

# 2. Build C++ application
cmake --preset MSVC_x64-debug
cmake --build --preset MSVC_x64-debug

# 3. Install Python dependencies
uv venv
uv pip install -r requirements.txt


# 4. Generate Python protobuf files for gRPC support
uv run python -m grpc_tools.protoc --python_out=./mcp_server --grpc_python_out=./mcp_server --proto_path=./proto ./proto/mcp_service.proto

⚠️ Note: After regenerating, you must manually check and update the import style at the top of mcp_server/mcp_service_pb2_grpc.py to:

from . import mcp_service_pb2 as mcp__service__pb2

Running the Server

Socket Mode (Default)

# Start C++ backend (socket server, default port 9876)
.\bin\cpp_app.exe socket 9876

# Start MCP server (in another terminal)
uv run mcp-server-demo --mode socket --socket-host localhost --socket-port 9876

gRPC Mode

# Start C++ backend (gRPC server, default 0.0.0.0:50051)
.\bin\cpp_app.exe grpc localhost:50051

# Start MCP server (in another terminal)
uv run mcp-server-demo --mode grpc --grpc-address localhost:50051

📋 Available Commands

Object Management

• create_object(name, type, properties): Create objects (cube, sphere, camera)
• delete_object(object_id): Delete objects by ID
• list_objects(): List all objects in the scene
• get_object_info(object_id): Get detailed object information

Software Operations

• get_software_info(): Get software information and version
• get_software_status(): Get current software status
• execute_software_command(command, params): Execute commands (render, clear_scene, reset_camera)

Project Management

• save_project(filename): Save current project to file
• load_project(filename): Load project from file

🏗️ Architecture

Communication Flow

AI Assistant ←→ Python MCP Server ←→（Socket or gRPC）←→ C++ Application

Command Format

Request:

{
    "type": "command_name",
    "params": {
        "param1": "value1",
        "param2": "value2"
    }
}

Response:

{
  "status": "success|error",
  "result": {...},
  "message": "error message if applicable"
}

🔧 Development

Adding New Commands

1. C++ Side: Add method to commandHandler class
2. Python Side: Add corresponding MCP tool function in server.py

📄 License

MIT License - Feel free to use and modify for your own projects.

🔗 References

• Model Context Protocol (MCP)
• FastMCP - Python MCP framework