comfyui-api

What This Skill Does

The ComfyUI API skill enables the OpenClaw AI agent to interface directly with a ComfyUI server instance via HTTP/WebSockets. It serves as a comprehensive bridge for automating complex generative workflows, including image, video, and audio production. The skill provides full control over the execution lifecycle: submitting JSON-formatted workflows, managing the server's task queue, uploading necessary assets (like reference images, models, or source audio), and performing capability probes to ensure the server is configured correctly. It features intelligent, automated timeout management—differentiating between lightweight image generation and heavy video rendering tasks—and simplifies the retrieval of complex outputs by handling file downloads automatically into your local directories.

Installation

To integrate this skill, use the ClawHub CLI tool in your environment: clawhub install openclaw/skills/skills/chsengni/comfyui-api-skill Ensure your system meets the requirements (Python 3.8+, requests, websocket-client, Pillow, and pyyaml). The skill assumes a default local connection at http://127.0.0.1:8188, though you can override this via the --server-url parameter if your ComfyUI instance is running on a remote GPU cluster.

Use Cases

This skill is perfect for automating batch image generation, converting static assets into animated video clips, processing audio-to-waveform visualizers, or managing long-running AI pipelines. Developers can use it to integrate ComfyUI into larger LLM-driven applications, allowing the agent to manage the state of the rendering queue without human intervention, while designers can leverage the workflow_manager to iterate through stored project configurations.

Example Prompts

1. "Run the image generation workflow 'fantasy_landscape.json' and save the result to my downloads folder."
2. "Check the current ComfyUI queue; if there are more than 5 tasks pending, let me know before starting a new video render."
3. "Upload 'character_sketch.png' as an input image and trigger the 'img2img_style_transfer' workflow using it."

Tips & Limitations

To get the most out of this skill, organize your workflows/ directory logically, as this allows the agent to recall and execute them by name instantly. If you are dealing with complex video animations (e.g., AnimateDiff or FrameInterpolation), allow for longer timeout periods. Note that the skill relies on the stability of your ComfyUI backend; ensure your custom nodes are installed on the server side prior to execution, as the skill merely facilitates the communication layer and does not install missing backend dependencies automatically.