Videocapture Mcp

Video Still Capture MCP is an MCP server that connects AI assistants to local webcams and video input devices using OpenCV. Designed for developers, computer vision experimenters, and desktop users using Model Context Protocol clients like Claude Desktop, the server enables language models to capture still photographs directly from connected hardware. Users can inspect physical objects, read printed text, and monitor physical environments through AI visual inspection. Despite the repository title, the server focuses strictly on capturing individual image frames rather than recording continuous video streams. It provides tools for both one-off quick snapshots and persistent camera sessions, allowing agents to capture single frames on demand without repeatedly reinitializing the device. Beyond basic capture, the server allows clients to query and adjust camera properties such as brightness, contrast, and resolution, as well as apply basic transformations like horizontal image flipping. Camera connections can be opened by device index, making it possible to target specific webcams on multi-camera setups, while automated resource management ensures hardware streams are cleanly closed when sessions end or when explicitly commanded.

Category: Design, Media & Creative

Tags: capture, media, recording, video

Visit Videocapture Mcp

How to install and configure Videocapture Mcp

1. Clone the repository and install dependencies in an environment with Python 3.10+: bash git clone https://github.com/13rac1/videocapture-mcp.git cd videocapture-mcp pip install -e . 2. Alternatively, register the server automatically using the MCP CLI: bash mcp install videocapture_mcp.py 3. To manually configure Claude Desktop, add the following entry to your claude_desktop_config.json: json { "mcpServers": { "VideoCapture": { "command": "uv", "args": [ "run", "--with", "mcp[cli]", "--with", "numpy", "--with", "opencv-python", "mcp", "run", "/ABSOLUTE_PATH/videocapture_mcp.py" ] } } } 4. Replace /ABSOLUTE_PATH/ with the absolute path to your local videocapture-mcp directory and restart Claude Desktop.

What you can do with Videocapture Mcp

• Take quick snapshots of real-world items or documents directly through Claude Desktop without managing ongoing camera connections. - Open persistent camera sessions across multi-step conversations to inspect objects from different angles using sequential single-frame captures. - Calibrate webcam hardware settings such as brightness, contrast, and resolution dynamically using the built-in video property tools. - Select specific cameras on multi-camera setups by providing custom device indices to capture views from alternate angles. - Mirror captured images with horizontal flipping to correct inverted camera orientations before sending them to the model for visual inspection.

Key facts

• https://github.com/13rac1/videocapture-mcp
• Design, Media & Creative
• capture, media, recording, video

Part of MCP Servers

What can Video Still Capture MCP do?

Video Still Capture MCP provides tools for AI models to access webcams via OpenCV. It allows assistants to perform instant single-frame captures, open and manage persistent camera connections, query and modify camera properties like brightness or resolution, and flip images horizontally for proper orientation.

Does Video Still Capture MCP record continuous video?

No. Despite the project name, Video Still Capture MCP only captures individual image frames from camera devices. It does not record, encode, or store continuous video streams or video files. AI clients use it exclusively to snap still photographs on demand.

Which MCP clients work with Video Still Capture MCP?

Video Still Capture MCP works with any client supporting the Model Context Protocol, including Claude Desktop on macOS, Linux, and Windows. It requires Python 3.10 or higher along with OpenCV and the MCP Python SDK installed on the host system.

How do I specify which webcam to use?

You can specify a webcam by passing the device index parameter to tools like quick_capture or open_camera. Index 0 corresponds to the primary default webcam, while index 1 or higher selects secondary external cameras connected to your system.

Is Video Still Capture MCP open source?

Yes, Video Still Capture MCP is open-source software released under the MIT License. You can view the source code, review camera interaction logic, report bugs, and submit pull requests directly through its public GitHub repository without proprietary software restrictions.