Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
v1.5.0: Standalone structured-prompt fix; 1920x1080 workflow; live-tested async JSON generation + delivery. Full changelog in SKILL.md.
What This Skill Does
Generates images via a locally running ComfyUI instance at localhost:8188 using the Flux2 workflow. Converts natural language descriptions into structured JSON prompts and submits them to the ComfyUI API. Supports async polling through a sub-agent that monitors generation every 5 seconds.
The async sub-agent mode lets the main agent continue other work during generation, using roughly 10x fewer tokens than blocking on the result.
When to use it
- Generating concept art from a text description
- Creating custom hero images for blog posts or presentations
- Prototyping visual ideas without switching to another tool
- Generating multiple scene variations with different seeds
- Auto-delivering finished images on completion without blocking
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Generate an image of a fluffy tabby cat sitting on a windowsill at sunset
- 1Converts the request to a structured JSON prompt with subjects, scene, lighting, mood, and camera fields
- 2Submits the prompt to ComfyUI via generate.py --submit-only and receives a prompt_id
- 3Spawns a watcher sub-agent that polls /history/{prompt_id} every 5 seconds
- 4Sub-agent downloads the completed JPG once generation finishes
- 5Sub-agent sends the image to the configured delivery target and removes the local file
1920x1080 JPG image generated via the Flux2 workflow and delivered to the configured channel
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.