Scumble — Claude Code MCP server: what it does, how to install
Connects AI coding assistants to the Scumble layered image editor via 116 MCP tools. It handles inpainting, prompt layout boxes, and layered PSD exports.
Open on GitHubWhat Scumble does
Scumble exposes its desktop editor capabilities to assistants through an MCP server. Agents can create selections, reference specific layers via @img1, inspect results, or apply GPU filter layers. Edits land as distinct layers rather than destructive pixel changes. The editor aligns edges automatically and lets models tweak color matching with a dedicated slider.
Rendering runs on local or remote ComfyUI nodes, Comfy Cloud, or third-party APIs like FLUX, fal.ai, and OpenAI. Without an active ComfyUI setup or provider keys, generation cannot run. The application currently supports Windows and Linux, while a macOS build remains unstarted.
What it helps with
- drive layered AI image inpainting workflows from an external agent
- guide generation by placing bounding boxes with separate descriptions
- export generated artwork into layered PSD or OpenRaster formats
- balance color blending between new edits and existing background layers
Example prompts for the agent
- Select the background on
@img1and inpaint a rainy cityscape using FLUX - Set up three prompt boxes on a blank canvas and generate the scene with ComfyUI
- Adjust edge matching on the active layer and export the document to PSD
How to install Scumble
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Scumble
How do you connect Scumble to Claude Code?
Launch the Scumble app, select *Help › Copy MCP registration*, and paste the resulting JSON configuration into your MCP client config file.
What does the Scumble MCP server do?
It provides 116 tools that let coding agents manage canvas layers, trigger inpainting recipes, and control document exports.
Is a local ComfyUI installation required?
No local installation is necessary. You can route jobs to Comfy Cloud or add external API credentials for providers like OpenAI, fal.ai, and Black Forest Labs.
