Image Toolkit MCP Server
@Kira-Pgr
Use AI to edit image in Claude Desktop / Cursor (AI P图)
Overview
What is Image Toolkit MCP Server?
An MCP server that transforms images using simple text commands, integrating Google Gemini for generation, Remove.bg for background removal, and FreeImage.host for hosting. Designed for users who want to edit photos with natural language.
How to use Image Toolkit MCP Server?
Install by cloning the repository, setting up a Python 3.11+ virtual environment with UV, and installing dependencies. Configure three API keys (Gemini, FreeImage.host, Remove.bg) and integrate with Claude Desktop or Cursor via JSON config or shell scripts.
Key features of Image Toolkit MCP Server
- Image generation from text prompts using Google Gemini
- Image modification based on text instructions
- Background removal via Remove.bg API
- Image hosting through FreeImage.host
- Resource management for generated and uploaded images
Use cases of Image Toolkit MCP Server
- Edit or transform images by describing changes in natural language
- Remove backgrounds from photos automatically
- Generate new images from text descriptions
- Host and share generated images via FreeImage.host
FAQ from Image Toolkit MCP Server
What are the runtime requirements?
Python 3.11 or higher and the UV package manager. Requires API keys for Google Gemini, FreeImage.host, and Remove.bg.
How do I integrate with Claude Desktop?
Add a server entry to your claude_desktop_config.json with the uv command pointing to server.py and your API keys in the env field.
How do I integrate with Cursor?
On Linux/macOS, modify cursor.sh with your API keys and project directory, then add an MCP server in Cursor settings with the command sh /absolute/path/to/cursor.sh. On Windows, use cursor.bat and set the command to cmd /c C:\absolute\path\to\cursor.bat.
What external services does this server depend on?
Google Gemini for image generation, Remove.bg for background removal, and FreeImage.host for image hosting.
Can I modify existing images with this server?
Yes. The server supports image modification based on text instructions, as shown in the demo (e.g., “add a coat to the dog”).
