Is the AI Toolkit Trainer skill free?

Yes. The skill itself is free from artokun/comfyui-mcp. SkillProof publishes the install command and an independent test verdict at no cost.

What is the AI Toolkit Trainer SkillProof Score?

7.2/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.

How do I install AI Toolkit Trainer?

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 AI Toolkit Trainer 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.

AI Toolkit Trainer

Train WAN 2.2 and Z-Image LoRAs with ostris AI-Toolkit: params, RunPod setup, ComfyUI use

by artokun · artokun/comfyui-mcp

Works with setup

Install — copy the command Download source ↗

Test report

Verdict: Works with setup
Score: 7.2/10
Tested: Jul 31, 2026
Environment: Claude Code 2.x (agent harness)
Upstream re-checked: Aug 10, 2026 · 3698855

Fetched SKILL.md via raw GitHub and cloned into an isolated HOME temp dir; frontmatter parses with name+description (plus a globs field), and no curl|sh/base64/eval smells are present in the skill. Spot-checked sibling packs it references: wan-t2v-video and anima-lora-trainer return 200, but z-image-base, z-image-turbo, and z-image-xy-plot all 404 (dangling references). Could not measure output: the deliverable (training a LoRA) needs the external ostris AI-Toolkit app, a training GPU, and a dataset, none runnable here, so any artifact would just restate the skill's own param tables. Docs are notably honest, with an explicit Unverified section flagging that the param tables are synthesized starting points, not read from the repo's config/examples.

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

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

What AI Toolkit Trainer does

A reference skill that guides training custom LoRAs with the standalone ostris AI-Toolkit trainer, covering WAN 2.2/2.1 (character, style, video-motion) and Z-Image (Turbo/Base) models. It provides local and RunPod install notes, dataset/caption prep, starting-point training parameters, VRAM guidance, and how to load the resulting .safetensors LoRA in a ComfyUI LoraLoaderModelOnly workflow. Triggers when the user wants to train a WAN or Z-Image LoRA.

How to install AI Toolkit Trainer

git clone --depth 1 https://github.com/artokun/comfyui-mcp.git /tmp/ai-toolkit-trainer-src
mkdir -p ~/.claude/skills
cp -R /tmp/ai-toolkit-trainer-src/plugin/skills/ai-toolkit-trainer ~/.claude/skills/ai-toolkit-trainer
# Self-contained knowledge skill: one SKILL.md, no local scripts or reference files.
# To actually TRAIN you must separately install ostris AI-Toolkit (https://github.com/ostris/ai-toolkit)
#   — a Node UI over a Python run.py backend — plus a CUDA training GPU and a captioned dataset.
#   The repo's Windows installer .bat files download launcher scripts from a third-party
#   HuggingFace repo (Aitrepreneur/FLX); review before running on a sensitive machine.
# Plugin-marketplace alternative (installs the whole comfyui-mcp plugin, 35 skills):
#   /plugin marketplace add artokun/comfyui-mcp  (repo ships .claude-plugin/marketplace.json)

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

Commands — how to trigger AI Toolkit Trainer

/ai-toolkit-trainer Train WAN 2.2 and Z-Image LoRAs with ostris AI-Toolkit: params, RunPod setup, ComfyUI use

It also activates on plain-language prompts like these:

Train a WAN LoRA on this video dataset
Set up a Z-Image LoRA on RunPod
Prepare my dataset for LoRA training

Frequently asked questions

Is the AI Toolkit Trainer skill free?: Yes. The skill itself is free from artokun/comfyui-mcp. SkillProof publishes the install command and an independent test verdict at no cost.
Does AI Toolkit Trainer work with Claude Code?: We tested it with Claude Code 2.x (agent harness) on Jul 31, 2026. Verdict: Works with setup. Fetched SKILL.md via raw GitHub and cloned into an isolated HOME temp dir; frontmatter parses with name+description (plus a globs field), and no curl|sh/base64/eval smells are present in the skill. Spot-checked sibling packs it references: wan-t2v-video and anima-lora-trainer return 200, but z-image-base, z-image-turbo, and z-image-xy-plot all 404 (dangling references). Could not measure output: the deliverable (training a LoRA) needs the external ostris AI-Toolkit app, a training GPU, and a dataset, none runnable here, so any artifact would just restate the skill's own param tables. Docs are notably honest, with an explicit Unverified section flagging that the param tables are synthesized starting points, not read from the repo's config/examples.
What is the AI Toolkit Trainer SkillProof Score?: 7.2/10 — installs cleanly 4/5, triggers reliably 5/5, output vs. baseline 5/10, docs & honesty 4/5.
How do I install AI Toolkit Trainer?: 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 AI Toolkit Trainer 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.