comfyui

Generate images, video, and audio by running ComfyUI workflows via REST and WebSocket APIs.

Updated Sep 9, 2026
One-click install
npx skills add https://github.com/luckybbjason1/trading --skill comfyui-luckybbjason1
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: comfyui
Source: https://github.com/luckybbjason1/trading/tree/main/.hermes/skills/creative/comfyui
Command: npx skills add https://github.com/luckybbjason1/trading --skill comfyui-luckybbjason1

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires comfy-cli, requests, websocket-client, and includes scripts (resource) and references (resource) components.

What problem does it solve? Running generative AI workflows in ComfyUI requires juggling installation, GPU compatibility, custom nodes, model downloads, and a low-level REST/WebSocket API. This Skill automates the full lifecycle — hardware checks, setup, dependency resolution, parameter injection, execution, and output download — for both local servers and Comfy Cloud. ## Core Features & Use Cases - Automated Setup & Health Checks: Probe GPU/VRAM, install ComfyUI via the official comfy-cli, launch the server, and verify readiness with a single health check script. - Workflow Execution with Parameter Injection: Inspect any API-format workflow to discover controllable parameters (prompt, seed, steps, dimensions), then run it with custom values, batch sweeps, and random seeds. - Dependency Management: Detect missing custom nodes, models, and embeddings in a workflow and auto-install them via comfy-cli. - Use Case: A user asks to generate 8 variations of an SDXL image with random seeds. The Skill checks the server, validates the workflow's dependencies, submits the batch with parallel execution, monitors progress over WebSocket, and downloads the resulting PNGs. ## Quick Start Ask the AI to generate an image with a prompt like "a sunset over mountains" using the ComfyUI skill, and it will check the server, pick a workflow, and run it for you.

Frequently Asked Questions about comfyui

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a ComfyUI workflow from the command line?

Use run_workflow.py with an API-format workflow JSON and a parameter dictionary, for example passing prompt, seed, and steps via --args. The script submits the job to the server, monitors execution, and downloads outputs to your chosen directory.

How do I install ComfyUI locally with comfy-cli?

Run the comfyui_setup.sh script, which checks your hardware, installs comfy-cli via pipx or uvx, installs ComfyUI with the right GPU flag, and launches the server on port 8188. Alternatively run comfy install --nvidia, --amd, --m-series, or --cpu manually.

What hardware do I need to run ComfyUI locally?

Local ComfyUI needs an NVIDIA GPU with at least 6 GB VRAM (8 GB for SDXL, 12 GB or more for Flux and video), an AMD GPU with ROCm on Linux, or an Apple Silicon Mac with at least 16 GB unified memory. Machines without a capable GPU should use Comfy Cloud instead.

Can I use Comfy Cloud instead of installing ComfyUI locally?

Yes, Comfy Cloud runs hosted workflows with an API key from platform.comfy.org set as COMFY_CLOUD_API_KEY. The scripts route requests to cloud endpoints automatically, though the free tier cannot run workflows via the API and requires a paid subscription.

Why does my ComfyUI workflow fail with class_type not found?

This error means a required custom node is not installed on the server. Run check_deps.py against the workflow to list missing nodes and models, then use auto_fix_deps.py or comfy node install to add the required packages.

What is the difference between API format and editor format workflows?

API format workflows have top-level node IDs with class_type fields and are directly executable via the REST API. Editor format contains nodes and links arrays for the visual UI and must be re-exported from ComfyUI using Workflow then Export (API) before scripts can run it.