Labsco
MCP SERVER

Changeimageto - Free Image Editing tools

by vipul510-web

Sixteen image operations behind one server — background removal, blur and grayscale, format conversion, upscale, OCR, inpainting, watermark cleanup, and bulk resize.

Image Analysis, OCR & Vision
Summary
Editing operations named the way you would ask for them.

The value of a wide tool list here is that each operation is its own call with its own name, so "remove the background and convert to WebP" maps onto two obvious steps rather than a parameter puzzle. Because the processing happens on the backend, nothing about your machine limits what you can run — and equally, the images leave it, which is the reason the API base is configurable. The size ceiling on inline results is the practical constraint: bulk jobs on large files are where you will want your own deployment.

What it is

A thin MCP client in front of the ChangeImageTo image API. The server itself does no image processing — it forwards to the hosted backend, or to your own deployment of the same routes, which means there is nothing heavy to install locally.

What you get
  • Background work: cut out the subject to a transparent PNG with a product or portrait mode, blur the background for a bokeh look, or desaturate only the background — `remove_background`, `blur_background`, `grayscale_background`
  • Colour grading across hue, saturation, brightness and contrast — `change_image_color`
  • Format conversion between PNG, JPEG, WebP and others — `convert_image_format`
  • Quality operations: upscale at 2×, 3× or 4×, and sharpen, denoise or deblur — `upscale_image`, `enhance_image`
  • Text handling in images: strip text out, replace text with an AI edit, or remove a Gemini watermark — `remove_text_from_image`, `edit_text_in_image`, `remove_gemini_watermark`
  • Extraction and export: OCR an image to text, or export it as a PDF — `image_to_text_ocr`, `image_to_pdf`
  • Inpainting against a mask you supply — `remove_painted_areas`
  • Batch operations over a list of images — resize, convert format, or run a quality check — `bulk_resize_images`, `bulk_convert_images`, `bulk_quality_check`
  • Three ways to hand over an image, so the source does not have to be moved first: an HTTPS URL, a local file path, or a data URL
Requirements

No account and no key against the default hosted backend — `CHANGEIMAGETO_API_BASE` points at it and can be repointed at your own deployment of the matching routes, with `CHANGEIMAGETO_API_TOKEN` supplying a bearer header if yours requires one. `CHANGEIMAGETO_HTTP_TIMEOUT` sets the HTTP timeout in seconds. Python 3.10+; clone and `pip install .`, or install straight from the git URL, then point your client at the server script or the installed console script. The package is `mcp-changeimageto` (0.1.0 in pyproject) and it speaks stdio. Binary results come back base64-encoded inside the JSON while the payload is small enough; larger outputs omit the base64 and say so, which is the case for pointing it at a local API or using the non-bulk tools instead.

Setup effort

One command — pip install git+https://github.com/vipul510-web/mcp-changeimageto.git