panel-operations

Provides on-demand procedures for operating the ComfyUI sidebar panel agent across graph editing tasks.

715|111|Updated Feb 15, 2026
One-click install
npx skills add https://github.com/artokun/comfyui-mcp --skill panel-operations
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: panel-operations
Source: https://github.com/artokun/comfyui-mcp/tree/main/plugin/skills/panel-operations
Command: npx skills add https://github.com/artokun/comfyui-mcp --skill panel-operations

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The ComfyUI panel agent's system prompt cannot hold every operational procedure, so multi-step tasks like subgraph refactoring, workflow merging, model downloading, and render debugging risk being done incorrectly or by guessing tool parameters.

Core Features & Use Cases

  • Graph Surgery Procedures: Step-by-step instructions for subgraphs, boundary rails, workflow merging via copy/paste, blueprint reuse, and untangling Get/Set-bus or rgthree toggle-heavy graphs.
  • Asset and Environment Operations: Covers CivitAI browsing, model weight downloads into correct ComfyUI subfolders, hardware stats, and Prompt Director audits.
  • Recovery and Debugging: Includes custom-node crash recovery, run-to-node render debugging, and multi-stage pipeline chaining with stage bypassing.
  • Use Case: When a render completes but looks wrong, follow the run-to-node procedure to bisect the graph with PreviewImage taps instead of re-running the whole workflow.

Quick Start

Ask the agent to debug a wrong render or refactor a large ComfyUI graph, and it will consult the relevant section of this skill before acting.

Frequently Asked Questions about panel-operations

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

FAQPage Schema
How do I merge two ComfyUI workflows into one graph?

Open the source workflow, select and copy the nodes, open the destination workflow, then paste. The clipboard survives the workflow switch, and the pasted nodes return new ids you wire and lay out yourself.

How do I debug a ComfyUI render that completes but looks wrong?

Use run-to-node partial execution: run only up to one output node, inspect its result, and bisect upstream until you find the first bad stage. Tap intermediate latents or masks with PreviewImage nodes to inspect non-output points.

Can I download model weights into ComfyUI without using curl or wget?

Yes, use the download_model tool with action download or download_civitai. It streams the file into the correct ComfyUI models subfolder and shows live progress in the panel's download tray, unlike raw shell downloads.

Why can't I add LoRA Manager autocomplete nodes in ComfyUI?

Nodes requiring AUTOCOMPLETE_TEXT_LORAS or AUTOCOMPLETE_TEXT_PROMPT inputs cannot be added programmatically; the add times out even when the pack is healthy. Use LoRA Text Loader (LoraManager) or the core LoraLoader node instead.

What are the limitations of editing ComfyUI subgraph boundary rails?

Removing a non-last boundary slot shifts positional host links, which can cause 'Required input is missing' errors on run. Repair by disconnecting and reconnecting remaining host links by name rather than by index.