gimp
GIMP MCP server
claude mcp add --transport stdio maorcc-gimp-mcp uv run --directory /full/path/to/gimp-mcp gimp_mcp_server.py \ --env GIMP_PATH="Path to the GIMP installation (used by the MCP server to locate GIMP binaries)" \ --env GIMP_VERSION="GIMP version (e.g., 3.0) if needed by plugins or scripts"
How to use
This MCP server provides a GIMP-integrated conversational interface. It exposes the full GIMP 3.0 API via PyGObject through the MCP protocol, enabling AI agents and other MCP clients to drive image editing operations with natural language. Core capabilities include image export and analysis, metadata retrieval, and the orchestration of multi-step workflows. Clients can request image bitmaps for viewing, pull image metadata without transferring heavy bitmap data, and perform complex edits by chaining GIMP operations. The server is designed to work with Claude Desktop, Gemini CLI, PydanticAI, and other MCP-compliant clients, allowing you to invoke GIMP features through a simple, language-agnostic interface.
How to install
Prerequisites:
- GIMP 3.0 or newer installed on your system
- Python 3.8+ installed
- uv (the Python MCP client/runner) installed
- An MCP-compatible client (e.g., Claude Desktop, Gemini CLI, PydanticAI)
Step-by-step installation:
- Clone the repository
git clone https://github.com/maorcc/gimp-mcp.git
cd gimp-mcp
- Install Python dependencies via uv
uv sync
- Ensure GIMP plugin is in place (follow the GIMP plugin installation instructions in the README):
- Copy gimp-mcp-plugin.py to your GIMP plug-ins directory and make it executable
- Restart GIMP
- Start the MCP server (from the gimp-mcp project directory)
uv run --directory /full/path/to/gimp-mcp gimp_mcp_server.py
- Connect with your MCP client (Claude Desktop, Gemini CLI, etc.) using the configuration example provided in the README, pointing to the above uv command.
Additional notes
Tips and common considerations:
- The server exposes an MCP-compliant image export and metadata tools; use get_image_bitmap() to fetch the current image as a base64-encoded PNG, and get_image_metadata() to retrieve image properties without transferring full bitmaps.
- Ensure the GIMP plug-ins directory is writable by the user running the MCP server, especially on macOS or sandboxed environments.
- If GIMP is launched remotely or via a headless setup, verify that the display environment (X11/Wayland) is properly configured for GIMP to run.
- The server typically starts at localhost:9877 as described in the Quick Start; configure your MCP client to point to that endpoint if required.
- If you modify the directory containing gimp_mcp_server.py, update the uv command accordingly in your MCP client setup.
Related MCP Servers
Gitingest
mcp server for gitingest
claude-ipc
AI-to-AI communication protocol for Claude, Gemini, and other AI assistants
davinci -professional
An enterprise-grade MCP server that exposes the full functionality of DaVinci Resolve and DaVinci Resolve Studio (through version 20) to either Claude Desktop or Cursor MCP clients. Fully configured and tested as a Claude Desktop Extension making installation as easy as clicking a button. Supports both Windows and Macintosh.
openapi -proxy
An MCP server that provides tools for exploring large OpenAPI schemas
microsoft
A Microsoft Graph API integration MCP server that provides AI assistants with comprehensive access to Microsoft 365 services through the Model Context Protocol (MCP). It enables natural language automation of email, calendar, file management, and contact operations across multiple Microsoft accounts.
zotero -lite
Zotero MCP Lite: Fast, Customizable & Light Zotero MCP server for AI research assistants