ComfyUI Agent Kit

Drives a local ComfyUI for image/video/audio generation and workflow building

Works with setup

Test report

Verdict
Works with setup
Score
7.6/10
Tested
Jul 17, 2026
Environment
Claude Code 2.x (agent harness)
Upstream re-checked
Aug 10, 2026 · d1dfb72

⚠ The author changed this skill after we tested it. The verdict below describes the version we ran on Jul 17, 2026; it's queued for a re-test.

The headline capability — driving a local ComfyUI install to generate images, video, or audio — could not be exercised here: nothing answered on :8188 and there is no GPU, so no generation delta over a no-skill baseline was measured. What is verifiable is real: the bundled workflow_layout.py took a 5-node graph piled at the origin (10 overlaps, 2 crossings) and laid it out left-to-right by dependency depth with zero overlaps and zero crossings. The knowledge base is genuinely deep, including a 112KB per-model prompting file, a node library, and hardware-fit logic. Scored setup rather than pass because the core value needs local hardware to demonstrate a beat over baseline.

Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works

  • Installs cleanly 5/5
  • Triggers reliably 5/5
  • Output vs. baseline 5/10
  • Docs & honesty 4/5

What ComfyUI Agent Kit does

A knowledge-plus-tooling skill for operating a local ComfyUI install over its HTTP API: per-model prompt recipes, the API/GUI workflow JSON format, hardware-aware model selection, VRAM coordination, and a graph auto-layout tool. Activates when generating images, video, or audio with ComfyUI, FLUX, LTX, Wan, Z-Image, or when building and running ComfyUI workflows. Requires a local ComfyUI server and GPU to do real work.

How to install ComfyUI Agent Kit

git clone https://github.com/SlavaSexton/ComfyUI-Agent-Kit
mkdir -p ~/.claude/skills
cd ComfyUI-Agent-Kit && cp -r claude-code/skills/comfyui ~/.claude/skills/comfyui

Skills live in ~/.claude/skills/ (global) or .claude/skills/ (per-project). Restart Claude Code after installing.

Commands — how to trigger ComfyUI Agent Kit

  • /comfyui Drives a local ComfyUI for image/video/audio generation and workflow building

It also activates on plain-language prompts like these:

  • generate a hero image with FLUX using my local ComfyUI install
  • build a text-to-video workflow graph in ComfyUI for me
  • my ComfyUI KSampler keeps outputting black images, help me fix it

Frequently asked questions

Is the ComfyUI Agent Kit skill free?
Yes. The skill itself is free from SlavaSexton/ComfyUI-Agent-Kit. SkillProof publishes the install command and an independent test verdict at no cost.
Does ComfyUI Agent Kit work with Claude Code?
We tested it with Claude Code 2.x (agent harness) on Jul 17, 2026. Verdict: Works with setup. The headline capability — driving a local ComfyUI install to generate images, video, or audio — could not be exercised here: nothing answered on :8188 and there is no GPU, so no generation delta over a no-skill baseline was measured. What is verifiable is real: the bundled workflow_layout.py took a 5-node graph piled at the origin (10 overlaps, 2 crossings) and laid it out left-to-right by dependency depth with zero overlaps and zero crossings. The knowledge base is genuinely deep, including a 112KB per-model prompting file, a node library, and hardware-fit logic. Scored setup rather than pass because the core value needs local hardware to demonstrate a beat over baseline.
What is the ComfyUI Agent Kit SkillProof Score?
7.6/10 — installs cleanly 5/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.
How do I install ComfyUI Agent Kit?
Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
Can I use ComfyUI Agent Kit with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.