
from comfyui_skills_openclaw263
Run and manage ComfyUI image-generation workflows via a CLI from AI agents—import, execute, check dependencies, and retrieve outputs across servers.
This skill exposes ComfyUI workflows to AI agents through a unified CLI. Agents can list available workflows, import new workflows, check and install node dependencies, submit image-generation jobs (txt2img/img2img/inpainting), and retrieve results. It bridges chat-based agents with ComfyUI servers for reproducible image generation.
Use when a user asks to generate images, run or import a ComfyUI workflow, perform img2img or inpainting tasks, check job status, or manage server/node dependencies. Ideal for agents assisting in AI-art generation or automating workflow execution across multiple servers.
Designed for agent runtimes that can execute shell commands and parse JSON outputs (OpenClaw, Claude Code, Codex, Hermes), especially those orchestrating image generation and workflow automation.
ComfyUI workflow orchestration skill for AI agents — import, execute, and manage image-generation workflows via a CLI. SKILL.md is well-structured with clear decision trees and command reference. All 3 scripts failed in sandbox: clean_generated_ui.sh needs project repo context, serve_pages.sh requires Ruby/bundle/Jekyll not available, update_frontend.sh needs gh CLI auth. No security red flags; minor concern about tarball download without checksum verification.
comfyui-skill-cliruby/bundle (jekyll)gh (GitHub CLI with GH_TOKEN)Scripts are operational/support tools (clean UI, serve docs, update frontend) rather than core skill functionality. The core skill works via the comfyui-skill CLI which isn't bundled. Minor shell quoting concerns in clean_generated_ui.sh and update_frontend.sh but not exploitable in context. update_frontend.sh downloads tar.gz from GitHub releases without checksum — standard practice for gh CLI but worth noting.