JM

jyjune/mcp_vms

Productivity workflow

Эта страница переведена автоматически. Технические детали смотрите в оригинале. Открыть оригинал

17 stars 7 forks Качество 69 Тренд 29MIT

MCP server for controlling and accessing a video surveillance system

Обзор

A Model Context Protocol (MCP) server designed to connect to a CCTV recording program (VMS) to retrieve recorded and live video streams. It also provides tools to control the VMS software, such as showing live or playback dialogs for specific channels at specified times. - Retrieve video channel information, including connection and recording status. - Fetch recording dates and times for specific channels. - Fetch live or recorded images from video channels. - Show live video streams or playback dialogs for specific channels and timestamps. - Control PTZ (Pan-Tilt-Zoom) cameras by moving them to preset positions. - Comprehensive error handling and logging. - Python 3.12+ - vmspy library (for VMS integration) - Pillow library (for image processing) If you want to use mcp-vms with Claude desktop, you need to set up the claude_desktop_config.json file as follows: The server uses the following default configuration for connecting to the VMS: - mcp_vms_config.

README

Показан фрагмент оригинального README. Открыть оригинал.

MCP Server - VMS Integration

A Model Context Protocol (MCP) server designed to connect to a CCTV recording program (VMS) to retrieve recorded and live video streams. It also provides tools to control the VMS software, such as showing live or playback dialogs for specific channels at specified times.

Features

  • Retrieve video channel information, including connection and recording status.
  • Fetch recording dates and times for specific channels.
  • Fetch live or recorded images from video channels.
  • Show live video streams or playback dialogs for specific channels and timestamps.
  • Control PTZ (Pan-Tilt-Zoom) cameras by moving them to preset positions.
  • Comprehensive error handling and logging.

Prerequisites

  • Python 3.12+
  • vmspy library (for VMS integration)
  • Pillow library (for image processing)

MCP-server Configuration

If you want to use mcp-vms with Claude desktop, you need to set up the claude_desktop_config.json file as follows:

{
  "mcpServers": {
	"vms": {
	  "command": "uv",
	  "args": [
		"--directory",
		"X:\\path\\to\\mcp-vms",
		"run",
		"mcp_vms.py"
	  ]
	}
  }
}

VMS Connection Configuration

The server uses the following default configuration for connecting to the VMS:

  • mcp_vms_config.py
vms_config = {
    'img_width': 320,
    'img_height': 240,
    'pixel_format': 'RGB',
    'url': '127.0.0.1',
    'port': 3300,
    'access_id': 'admin',
    'access_pw': 'admin',
}

Installation

1. Install UV Package Manager

Run the following command in PowerShell to install UV:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

For alternative installation methods, see the official UV documentation.

2.Install VMS Server

Download and install the VMS server from:
http://surveillance-logic.com/en/download.html (Required before using this MCP server)

3.Install Python Dependencies

Download the vmspy library:
vmspy1.4-python3.12-x64.zip Extract the contents into your mcp_vms directory

The mcp-vms directory should look like this:

mcp-vms/
├── .gitignore
├── .python-version
├── LICENSE
├── README.md
├── pyproject.toml
├── uv.lock
├── mcp_vms.py            # Main server implementation
├── mcp_vms_config.py     # VMS connection configuration
├── vmspy.pyd             # VMS Python library
├── avcodec-61.dll        # FFmpeg libraries
├── avutil-59.dll
Открыть полный README на GitHub

Установка

This server does not publish a one-line install command.

Open the repository installation guide Новичок в MCP? Начните с руководства на mcpskill.org →

Конфигурация

{ "mcpServers": { "vms": { "command": "uv", "args": [ "--directory", "X:\\path\\to\\mcp-vms", "run", "mcp_vms.py" ] } } }