ImageSorcery MCP

What it does

ImageSorcery MCP connects your AI assistant to a powerful suite of local image processing capabilities using OpenCV and Ultralytics. It allows the AI to perform complex image manipulations, analyze content, and extract text directly on your local machine without uploading images to external servers.

Tools

• blur: Blurs specified areas of an image.
• change_color: Changes the color palette (e.g., to sepia).
• crop: Precise image cropping via coordinates.
• detect: Object detection using YOLO models.
• draw_arrows/draw_circles/draw_lines/draw_rectangles/draw_texts: Visual annotations on images.
• fill: Fills areas with color or removes backgrounds.
• find: Finds objects based on text descriptions (CLIP).
• get_metainfo: Extracts image metadata.
• ocr: Optical Character Recognition via EasyOCR.
• overlay: Overlays images with transparency.
• resize: Changes image dimensions.
• rotate: Rotates images with boundary handling.

Installation

Install via pipx:

pipx install imagesorcery-mcp
imagesorcery-mcp --post-install

Add to claude_desktop_config.json:

"imagesorcery-mcp": {
  "command": "imagesorcery-mcp",
  "timeout": 100
}

Supported hosts

Confirmed support for Claude Desktop, Cursor, and Cline.