Diagnostics

First thing you should do after installing AI Fashion Studio: drop this node on an empty canvas and run it. It takes no inputs, calls no API, and spends nothing. What you get back is one JSON blob that tells you the pack actually loaded, your configuration parses, and whether a key is waiting in the environment. When something downstream goes wrong, this is where you'll wish you had started.

AI Fashion Studio is an API-first pack: it does almost nothing useful until it talks to a provider, and its whole setup story is environment variables, not model downloads. That means most install failures are silent configuration failures - and this node exists to make them loud. Under the hood it calls build_diagnostics(), which loads your settings through Settings.from_env(). That's the important part: reading the environment validates it. If an AIFS_* variable is set to something nonsense, this node throws a ConfigurationError naming the culprit, before you ever waste a paid call discovering it.

The output is a single string, diagnostics_json. Preview it with any text-display node or just look at the node's output. It includes the package version, the milestone, the pinned ComfyUI API version (v0_0_2), and a safe summary of your settings - output root, candidate ceiling, cost ceiling, the OpenAI model names. Crucially, it reports openai_configured as a plain boolean. It never prints the key itself (secrets_included: false), just whether one is set. That's the right behavior for a pack whose whole security posture is "secrets live in the process environment, never in the workflow."

Installation is the same as any custom node - ComfyUI Manager, search "AI Fashion Studio", or:

cd ComfyUI/custom_nodes
git clone https://github.com/KillPhantom/ai-fashion-studio
cd ai-fashion-studio && pip install -r requirements.txt
# restart ComfyUI

One heads-up specific to this pack: it's written against ComfyUI's newer V3 backend (comfy_api.v0_0_2, io.ComfyNode, io.Schema), and it pins requires-comfyui >=0.28.0. If you're on an older ComfyUI, the nodes simply won't register. Also, if you open the source and see no NODE_CLASS_MAPPINGS dictionary, that's not broken - it's the new style.

Where people get burned: the README's .env.example is a reference, not something ComfyUI auto-loads. Set your variables in the launch environment or you'll get defaults, and the diagnostics node is the only place you'll see what actually got picked up. The pack is also brand new - no community track record yet - so reading the source before you hand it a key is cheap insurance. Run this node first, read the JSON, then go make a mess elsewhere.

It's a smoke test, not a feature. But it's the single most useful node in the pack on day one.