šŖ ImageSorcery MCP ā an MCP server that equips AI assistants with image-processing tools (crop, resize, detect objects, OCR, etc.)
https://github.com/sunriseapps/imagesorcery-mcpYour AI assistant sees an image and describes it beautifully. Ask it to crop that image to focus on the subject? "I cannot modify images directly." Want it to extract text from a form or detect objects in photos? More limitations.
ImageSorcery MCP flips this completely. Your AI suddenly becomes a computer vision powerhouse with 13+ professional image processing tools at its disposal.
Instead of working around image limitations, your AI can:
Process images with precision: Crop, resize, rotate with pixel-perfect control using OpenCV under the hood
Extract information intelligently: OCR text from documents, detect objects with YOLO models, analyze image metadata
Edit images contextually: Add text overlays, draw annotations, apply blur effects, change color palettes
Understand visual content: Use CLIP models to find objects based on natural language descriptions ("find all the dogs in this folder")
Here's what changes in practice:
You: "Copy all photos with pets from my 'vacation' folder to 'pets' folder"
AI: *analyzes each image, detects animals, moves relevant files*
You: "Find the cat in photo.jpg and crop the image to center it"
AI: *detects cat location, calculates crop coordinates, saves centered result*
You: "Extract and format all text from these form images"
AI: *runs OCR, structures the data, creates organized output*
ImageSorcery runs as an MCP server that plugs directly into Claude, Cursor, Cline, or any MCP-compatible AI client. When your AI needs to process images, it calls these tools seamlessly.
Core image operations: resize, crop, rotate, blur, color manipulation, overlay composition
Computer vision: YOLO object detection, CLIP-based semantic search, EasyOCR text extraction
Professional features: Metadata extraction, batch processing, format conversion, annotation tools
The server handles the heavy lifting ā model loading, image processing pipelines, file management ā while your AI focuses on understanding what you want accomplished.
No complex APIs to learn. Install the package, add a few lines to your MCP configuration, and your AI immediately gains these capabilities:
pip install imagesorcery-mcp
imagesorcery-mcp --post-install # Downloads required models
Configuration is minimal:
"mcpServers": {
"imagesorcery-mcp": {
"command": "imagesorcery-mcp",
"timeout": 100
}
}
Your AI can now say things like "I'll analyze the images in your folder and extract text from each form" instead of "I can't process image files."
Document processing: Your AI scans receipts, forms, and documents, extracting structured data without manual typing
Photo organization: Automatically sort photos by content, create albums based on detected objects or people
Content creation: Generate annotated images, add branded overlays, batch process marketing materials
Data extraction: Pull information from screenshots, charts, diagrams that would otherwise require manual transcription
ImageSorcery uses production-ready libraries: OpenCV for image processing, Ultralytics for object detection, EasyOCR for text recognition. The models are industry-standard (YOLOv11, CLIP) with proven accuracy.
Multiple deployment modes support everything from local development to production servers. The FastMCP framework handles the protocol complexity, letting you focus on what your AI should accomplish with images.
This isn't another demo that breaks in production. It's built to handle real workloads where image processing becomes a natural part of your AI workflows.
Your AI assistant just gained professional computer vision capabilities. Time to put them to work.