thegmpteam/thegmpteamstudios repository overview

⁠ComfyUI MCP — TheGMPTeamStudios HUD Docker Image

This Docker image provides a fully integrated ComfyUI MCP (Master Control Program) server with a Neon-style HUD frontend, ready to run. It allows you to generate AI images in real-time using ComfyUI workflows, directly from a visually rich interface.

⁠Key Features

• WebSocket API for generating images, listing workflows, and retrieving available models.
• HTTP server serving a single-page Neon HUD (index.html) with live logs, image preview, and workflow/model selection.
• Dynamic workflow management: supports custom JSON workflows for flexible image generation.
• Model selection: dynamically lists all available ComfyUI models.
• Persistent image history stored in the browser’s localStorage.
• Optional starting image: provide an initial image to use as input for image-to-image generation workflows.
• Neon-style interface: cyan glow accents, animated log messages, and responsive layout.
• Ready for containerized deployment: exposes ports 9000 (WebSocket) and 9001 (HTTP).

---

⁠Running the Docker Image

You can run the container with options to customize the host, ports, workflows directory, and starting image:

docker run -d --name ComfyUIMCP --gpus all -p 9001:9001 -p 9000:9000 -e COMFYUI_HOST=host.docker.internal -e COMFYUI_PORT=8188 -e HTTP_PORT=9001 -v D:\MCP\Workflows:/app/workflows --restart always thegmpteam/thegmpteamstudios:latest

docker run -d
-p 9000:9000 \ # WebSocket port -p 9001:9001 \ # HTTP HUD port -v /local/path/workflows:/app/workflows \ # Mount workflow JSON files -e COMFYUI_HOST=<comfyui_host> \ # Default: localhost -e COMFYUI_PORT=<comfyui_port> \ # Default: 8188 -e WORKFLOWS_DIR=/app/workflows
-e MCP_HOST=0.0.0.0
-e WS_PORT=9000
-e HTTP_PORT=9001
--name ComfyUIMCP
thegmpteam/thegmpteamstudios:latest

Environment Variables Variable Default Description COMFYUI_HOST localhost Host where ComfyUI server is running COMFYUI_PORT 8188 ComfyUI API port WORKFLOWS_DIR /app/workflows Path to workflow JSON files inside the container MCP_HOST 0.0.0.0 Host for MCP servers WS_PORT 9000 WebSocket port for MCP HTTP_PORT 9001 HTTP port for HUD

Starting Image Option The Neon HUD supports optional starting images for image-to-image workflows.

You can upload an initial image through the HUD when generating images.

Accessing the HUD Open a browser and navigate to:

cpp

http://<docker_host>:9001/ Left panel: Enter prompts, select workflows and models, optionally upload a starting image, and generate images.

Right panel: Preview generated images, download them, and browse your persistent image history.

WebSocket API WebSocket endpoint: ws://<docker_host>:9000

Supported tools:

generate_image: Generate an image via a workflow and model

list_workflows: List available workflow JSON files

list_models: List available ComfyUI models

Example JSON request:

json { "tool": "generate_image", "params": { "prompt": "A futuristic cityscape", "workflow_id": "basic_api_test", "width": 512, "height": 512, "model": "v1-5-pruned-emaonly.ckpt" } }

Workflow Example workflows/basic_api_test.json:

json

{ "node1": { "type": "CLIPTextEncode", "inputs": { "text": "" } }, "node2": { "type": "KSampler", "inputs": { "seed": 0 } }, "node3": { "type": "Resolution", "inputs": { "width": 512, "height": 512 } }, "node4": { "type": "Model", "inputs": { "model_name": "v1-5-pruned-emaonly.ckpt" } } }

Notes Ensure ComfyUI server is running and accessible at COMFYUI_HOST:COMFYUI_PORT.

History of generated images is stored locally in the browser.

The Neon HUD interface is fully responsive with animated logs and supports multiple screen sizes.

Works for artists, developers, and AI enthusiasts who want a visually immersive interface to control ComfyUI workflows.