ComfyUI Flux

Provides a self-contained MCP server for generating images with FLUX models via ComfyUI, including GPU-accelerated execution and Claude Desktop integration.

typescript

3

GitHub Stars

typescript

Language

5 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "dhofheinz-mcp-comfyui-flux": {
      "command": "wsl.exe",
      "args": [
        "bash",
        "-c",
        "cd /path/to/mcp-comfyui-flux && docker exec -i mcp-comfyui-flux-mcp-server-1 node /app/src/index.js"
      ]
    }
  }
}

You run a self-contained MCP (Model Context Protocol) server that orchestrates image generation with FLUX models via ComfyUI. It connects your Claude Desktop workflow to a streamlined, GPU-accelerated container stack, giving you fast, reliable image creation with built-in upscaling, background removal, and model management—all locally on your machine.

How to use

You interact with the MCP server through your MCP client (such as Claude Desktop) to request image generation, upscaling, and background removal. The server runs in a local container environment and exposes a WebSocket-based API that the client uses to send prompts, configure generation parameters, and receive results. You can start with a quick image generation flow, then add upscaling and background removal as part of your post-processing pipeline.

Key capabilities you can leverage include generating images with FLUX schnell fp8 for fast previews, performing 4x upscaling with UltraSharp or AnimeSharp models, and removing backgrounds with RMBG-2.0. You can also generate batches of images in parallel and switch between fast Schnell and higher-quality Dev configurations as needed.

How to install

Prerequisites you need to prepare before installation:

Docker installed (20.10+).
NVIDIA CUDA toolkit and NVIDIA Container Toolkit if you plan GPU acceleration (required for GPU workflows).
A capable machine with 16 GB RAM or more (32 GB recommended for GPU workloads).
Optional: Claude Desktop for MCP client integration.

Install and configure the MCP Flux environment using the provided installer or build script. Follow these steps exactly to set up the system and start services.

# Standard installation
./install.sh

# Non-interactive installation
./install.sh --yes

# CPU-only mode
./install.sh --cpu-only

# With specific models (minimal/all/none/auto)
./install.sh --models minimal

# Debug mode
./install.sh --debug

Build and start flow

If you prefer to build manually, use the optimized build flow to create the containers and start services.

# Build only
./build.sh

# Build and start
./build.sh --start

# Build with cleanup
./build.sh --start --cleanup

# Rebuild without cache
./build.sh --no-cache

Available tools

generate_image

Generate images using FLUX schnell fp8 with optimized defaults. Supports prompt, size, steps, seed, and batch options.

upscale_image

Upscale generated images to 4x using ultrasharp or animesharp models with optional content type handling.

remove_background

Remove the image background using RMBG-2.0 with alpha matting and selectable output format.

check_models

Verify available models in ComfyUI to ensure required assets are present.

connect_comfyui

Manage the ComfyUI connection so that the MCP server can communicate with the UI layer.