gimp-mcp
Control GIMP from AI assistants via the Model Context Protocol, enabling image editing automation through natural language commands.
README
GIMP MCP Server
Control GIMP from Claude Desktop, Codex, Cursor, and other Model Context Protocol clients. This project provides a local MCP server and a GIMP 3.x Python plug-in that expose GIMP automation through a secure Unix socket, including Procedure Database discovery, image export, canvas snapshots, context inspection, and batch operations.
Keywords: GIMP MCP, GIMP Model Context Protocol, Claude GIMP, Codex GIMP, AI image editing, GIMP automation, GIMP Python plug-in, GIMP PDB, MCP server.
Why This Exists
GIMP has a powerful Python and Procedure Database API, but it is not easy for AI assistants to use directly. GIMP MCP bridges that gap:
- Ask an AI assistant to inspect or edit images in GIMP.
- Search and call GIMP PDB procedures from MCP tools.
- Capture the current canvas as MCP image content for visual verification.
- Run repeatable image workflows without clicking through the GIMP UI.
- Keep automation local by default with a user-owned Unix socket.
The normal MCP path auto-starts a private headless GIMP bridge, so users do not
need to open GIMP or click Filters > Development.
Features
- GIMP 3.x automation through MCP
- PDB search, describe, and call for broad GIMP feature coverage
- MCP image capture with optional crop and max-size scaling
- Open image and layer inspection
- First-class layer tools for listing and updating layer properties
- Friendly drawing tools for rectangles, ellipses, text, and polylines
- Foreground/background, brush, font, opacity, antialias, and feather state
- File open/export helpers for PNG, JPEG, WebP, TIFF, GIF, BMP, and XCF
- Batch bridge operations
- Headless auto-start for production-style automation
- Manual visible-GIMP bridge for debugging interactive sessions
- macOS-friendly installer for
/Applications/GIMP.app
Requirements
- GIMP 3.x installed locally
- Python 3.11+
- An MCP-compatible client such as Codex, Claude Desktop, Cursor, or another stdio MCP client
This repo defaults to macOS and /Applications/GIMP.app. The bridge itself is
Python/GIMP based and can be adapted for Linux by installing the plug-in into
the relevant GIMP profile directory.
Quick Start
git clone https://github.com/abelduarte/gimp-mcp.git
cd gimp-mcp
python3 -m venv .venv
. .venv/bin/activate
python -m pip install -e .
./scripts/install_bridge.sh
./scripts/smoke_test.sh
If GIMP is not installed at /Applications/GIMP.app:
GIMP_APP=/path/to/GIMP.app ./scripts/install_bridge.sh
GIMP_APP=/path/to/GIMP.app ./scripts/smoke_test.sh
MCP Client Configuration
Use the wrapper script as your MCP server command:
/absolute/path/to/gimp-mcp/scripts/run_mcp_server.sh
The wrapper starts the stdio MCP server and automatically launches a private headless GIMP bridge if one is not already running.
Codex
Add this to your Codex MCP config:
[mcp_servers.gimp]
command = "/absolute/path/to/gimp-mcp/scripts/run_mcp_server.sh"
Claude Desktop
Add this to your Claude Desktop config:
{
"mcpServers": {
"gimp": {
"command": "/absolute/path/to/gimp-mcp/scripts/run_mcp_server.sh"
}
}
}
Project-local MCP discovery
The repository includes .mcp.json for clients that support project-local MCP
server discovery.
How It Works
There are two pieces:
gimp_mcp.server: the stdio MCP server used by the AI client.bridge/gimp_mcp_bridge.py: a GIMP Python plug-in that runs inside GIMP and exposes a JSON bridge over a local Unix socket.
Default socket:
/tmp/gimp-mcp-$USER.sock
The socket is created with 0600 permissions. Anyone who can connect to it can
drive the GIMP process, so do not expose it to untrusted users.
Available Tools
gimp.status: check bridge healthgimp.get_info: inspect GIMP version, directories, session, platform, and PDBgimp.get_context_state: inspect colors, brush, opacity, font, antialias, and feathergimp.get_image_bitmap: return a GIMP image as MCP image contentgimp.list_images: list open images and layer treesgimp.get_image: inspect one image by IDgimp.list_layers: list layers for one imagegimp.set_layer_properties: rename layers or update visibility, opacity, and blend modegimp.procedure.search: search GIMP's Procedure Databasegimp.procedure.describe: inspect procedure arguments and return valuesgimp.procedure.call: call a GIMP PDB proceduregimp.open: open an image filegimp.export: export an image filegimp.fill_rect: fill a rectangle on a drawablegimp.fill_ellipse: fill an ellipse on a drawablegimp.add_text: create a text layergimp.draw_polyline: draw deterministic pixel polylinesgimp.batch: run bridge operations in sequence
Example PDB Call
Search before calling:
{
"query": "image-resize",
"limit": 10
}
Then call the procedure with exact PDB argument names:
{
"name": "gimp-image-resize",
"args": {
"image": { "id": 1, "type": "image" },
"new-width": 1920,
"new-height": 1080,
"offx": 0,
"offy": 0
}
}
Visual Verification
Use gimp.get_image_bitmap to let the AI client see the current GIMP image:
{
"image_id": 1,
"max_width": 1024,
"max_height": 1024
}
Crop a region for faster iteration:
{
"image_id": 1,
"region": {
"origin_x": 100,
"origin_y": 100,
"width": 400,
"height": 300
},
"max_width": 800,
"max_height": 600
}
Manual Visible-GIMP Mode
For normal automation, use scripts/run_mcp_server.sh. It auto-starts a
headless bridge.
For debugging against the visible GIMP app:
- Open GIMP.
- Run
Filters > Development > Start MCP Bridge. - Start your MCP client with the same
GIMP_MCP_SOCKET.
This manual menu is a debug fallback, not the recommended production path.
Development
Run the smoke test:
./scripts/smoke_test.sh
Run full local diagnostics:
./scripts/doctor.sh
Compile-check the Python files:
python3 -m py_compile src/gimp_mcp/server.py bridge/gimp_mcp_bridge.py
Install the bridge after editing it:
./scripts/install_bridge.sh
Examples
Generated examples are in examples/generated/, including:
horse-from-gimp-mcp.pngbar-chart-from-gimp-mcp.pngshapes-text-demo.png
Security
GIMP MCP is a local automation bridge. Treat access to the Unix socket as access to the running GIMP process.
- Do not expose the socket to untrusted users.
- Do not run the bridge as root.
- Prefer the default Unix socket transport over TCP.
- Review
SECURITY.mdbefore enabling any future raw Python execution feature.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.