comfy-ai
by tullyhu
Web & Frontend DevelopmentBrief
Integrates with local ComfyUI instance at http://192.168.31.7:8000.
Description
What This Skill Does
Connects to a local ComfyUI instance to generate images from text prompts or edit uploaded images using a reference workflow. All processing runs on the user's machine using the Flux2-Klein-4B model with VAE and CLIP. Returns generated images directly as chat attachments.
All image generation and editing happens locally, so no prompts or images are sent to external services.
When to Use It
- Generating concept art from a text description
- Editing a product photo with a new visual style
- Adding effects or branding elements to an uploaded image
- Drafting visual mockups without leaving the chat interface
- Prototyping image variations from a reference photo
Example Workflow
User asks: Edit this image [uploads handbag_white.png] to add neon glow and a logo
- 1
Detects uploaded image in the message
- 2
Copies image to ~/clawd/skills/comfy-ai/input/
- 3
Loads the img2img workflow JSON
- 4
Sends request to local ComfyUI at http://192.168.31.7:8000
- 5
Retrieves generated image from the output folder
Edited image returned as a media attachment in chat
Requirements
- ComfyUI installed and running at http://192.168.31.7:8000
- Workflow file at ~/clawd/skills/comfy-ai/image_flux2_klein_image_edit_4b_distilled.json
Installation
clawhub install tullyhu/comfy-aiRun the CLI command, or paste the prompt into Claude Code, Cursor, Codex or whichever coding agent you use.