MCP Web Cam Server

What is MCP Web Cam Server?

MCP Web Cam Server is a Model Context Protocol (MCP) server that enables LLMs to control webcams—capturing photos, recording videos, and managing camera settings—through the MCP protocol. It is designed for use with MCP‑compatible clients like Claude Desktop.

How to use MCP Web Cam Server?

Install the server as an MCP server using npx -y @kmizu/mcp-web-cam or by cloning the repo and building manually. For Claude Desktop, add the corresponding configuration to claude_desktop_config.json and restart the app. Other MCP clients can run the server via node /path/to/dist/index.js using stdio transport. Tools such as select_camera, capture_photo, start_recording, and list_cameras are invoked by the client.

Key features of MCP Web Cam Server

• Capture photos with adjustable width, height, quality, and format
• Record video with configurable duration, fps, format, and codec
• List available cameras and display a live Web UI for selection
• Adjust camera settings like brightness, contrast, saturation, and focus
• Cross‑platform support for Linux, Windows, and macOS
• Resources provide access to captured photos and recordings

Use cases of MCP Web Cam Server

• An LLM powered by MCP captures a photo from a user’s webcam on demand
• A chatbot records a short video clip with specified duration and codec
• A smart assistant selects the correct camera and tweaks exposure before capturing
• An automation workflow lists all connected cameras and returns their details

FAQ from MCP Web Cam Server

What are the system dependencies?

Linux requires v4l-utils and ffmpeg. macOS needs ffmpeg (via Homebrew). Windows requires FFmpeg and Windows Media Foundation.

How do I install MCP Web Cam Server for Claude Desktop?

Use the npx method—add { "mcpServers": { "webcam": { "command": "npx", "args": ["-y", "@kmizu/mcp-web-cam"] } } } to your Claude Desktop configuration file, then restart the app.

What tools does the server provide?

Tools include select_camera, get_current_camera, capture_photo, start_recording, stop_recording, list_cameras, get_camera_settings, and set_camera_settings. Each tool has optional parameters documented in the README.

How are captured photos returned?

The capture_photo tool supports a return_type parameter that can be 'location' (file path), 'buffer', or 'base64'.

Where are captures and recordings stored?

Photos are saved to a captures/ directory and videos to a recordings/ directory, both auto‑created in the server’s working folder.