Andonstar MCP Microscope Controller

This project provides a Model Context Protocol (MCP) implementation for Andonstar digital microscopes (optimized for the AD207S). It enables AI assistants, such as Claude, to interact with your microscope for electronics repair, inspection, and documentation.

Features

• Model Context Protocol Integration: Seamlessly connects your microscope to LLM-based workflows.
• Smart Hardware Discovery: Automatically scans USB ports to find the high-resolution microscope device.
• High-Definition Capture: Reliable 1920x1080 MJPG frame grabbing with sensor stabilization.
• Interactive Focus Assistant: A real-time sharpness scoring window with peak value tracking.
• Project-Based Organization: Automatic watermarking and folder-based sorting for your repair captures.
• Audio Feedback: System beeps to notify the user of video recording states.
• AI-Powered Annotation: Captures a fresh image and lets AI place numbered markers on exact coordinates — no manual path handling required.

Project Structure

Andonstar-MCP/
├── captures/               # Snapshot and video storage (git-ignored)
├── src/
│   ├── server.py           # FastMCP Wrapper (Main Entry Point)
│   ├── vision_engine.py    # Core camera and image processing logic
│   └── tools/              # Diagnostic and research scripts
│       ├── check_capabilities.py
│       └── debug_hd_capture.py
├── .gitignore              # Rules for Git exclusion
└── README.md               # You are here

Installation

1. Prerequisites

• Python 3.10 or higher
• A digital microscope (e.g., Andonstar AD207S) connected via USB.

2. Install Dependencies

pip install opencv-python mcp

3. Setup Claude Desktop

Add the following configuration to your claude_desktop_config.json (usually located in %APPDATA%\Claude\ on Windows):

{
  "mcpServers": {
    "andonstar": {
      "command": "python",
      "args": ["C:/Users/YOUR_USER/Path/To/Project/src/microscope_server.py"]
    }
  }
}

Usage

Once the server is running, you can ask Claude to:

• "Capture an image of this PCB."
• "Set the project name to Sony TC-FX44."
• "Start the focus assistant."
• "Record a 10-second video of the soldering process."

Diagnostic Tools

If you encounter issues with resolution or camera detection, use the provided tools in src/tools/:

• check_capabilities.py: Verifies if the OS can see the 1080p stream and displays the FOURCC code.

License

This project is licensed under the MIT License.