ChatGPT outfit prompts that use your own wardrobe
Nine ChatGPT outfit prompts to copy, built around what OpenAI says the model struggles with, plus how to keep it working from clothes you really own.

The best ChatGPT outfit prompts give it a written list of your clothes, the day’s plans and the weather, then tell it to use only what’s on the list. Without a list, it has no way of knowing what you own. Below are nine prompts to copy, and each one is shaped around limits OpenAI itself has published.
A note on what this is. We wrote these prompts and based them on OpenAI’s documentation, checked on 8 November 2026. We haven’t scored ChatGPT’s answers, so we won’t tell you how good they’ll be. Try one and judge for yourself.
Start by writing down what you own
Everything else depends on this step. ChatGPT can’t see your wardrobe, so it only knows what you tell it.
You can photograph clothes for this, and OpenAI’s image FAQ says image inputs work on free and paid plans. The same page lists limits that shape how to do it:
| What OpenAI says | What to do about it |
|---|---|
| Counts of objects may be approximate | Don’t photograph a packed rail. Add a few pieces at a time |
| Descriptions can be wrong | Read the list back and correct it |
| Rotated images may be misread | Keep photos upright |
| Unclear images give less accurate results | Use daylight and a plain background |
| File names and metadata aren’t processed | Calling a photo “navy-wool-coat” won’t help. Say it in the message |
| The number of images per chat depends on size and text | If uploads fail, send fewer or smaller images |
Then ask for an inventory you can keep:
Prompt 1: build the list. Here are photos of clothes I own. List each item on its own line with category, colour, fabric if you can tell, and how formal it is from 1 to 5. If you aren’t sure about something, write “unsure” instead of guessing. Don’t add anything that isn’t in the photos.
Check the answer against the real clothes. A navy jumper called black will give you bad colour advice for as long as you use the list. Save the corrected version in your notes.

Why you should paste the list every time
ChatGPT’s Memory can help, but OpenAI says it doesn’t keep every detail, and it decides for itself what’s relevant. A wardrobe of 80 items is exactly the sort of detail that can get lost or go out of date. Pasting your list at the start of a chat takes a few seconds and removes the doubt.
Tell it about changes as well. If you’ve given away the grey coat, delete it from your notes. It’s easy to end up with a tidy plan that includes something you no longer have.
Eight more prompts to copy
Replace the parts in square brackets. Paste your list underneath each prompt.
Prompt 2: today’s outfit. Using only the items in my list, suggest three outfits for [plans] today. It’s [temperature] and [dry, windy or raining], and I’ll be [walking, sitting, outdoors]. For each outfit, name every piece and say in one line why it fits the day. Don’t suggest anything that isn’t on the list.
Prompt 3: the piece I never wear. I own [item] and hardly ever wear it. Using only my list, give me three outfits built around it, from most casual to smartest. Then tell me what might be stopping me from wearing it, as questions I can answer.
Prompt 4: more from less. Choose 10 items from my list that combine with each other as much as possible. Show how many outfits they make and list them. Tell me which item appears least and whether I could replace it with something else I own.
Prompt 5: dress code check. The invitation says [exact wording]. It’s at [place and time]. Tell me what that wording usually means and pick the closest outfit from my list. If nothing fits, say so instead of stretching the rules.
Prompt 6: gap check before buying. I’m thinking of buying [item, price]. Using my list, name the outfits it would add and which pieces it would duplicate. Be sceptical: argue for not buying it first, then for buying it.
Prompt 7: a week of outfits. Plan [number] outfits for next week from my list. My days are [plans]. Don’t repeat a top within the week, and keep to what can be washed by [day]. Flag anything that needs ironing or dry cleaning.
Prompt 8: packing. I’m going to [place] for [days], with [luggage]. The forecast is [details from a weather site]. Choose the fewest items from my list that will cover [plans]. Show the outfits by day and what I’ll wear while travelling.
Prompt 9: second opinion on a photo. This is what I’m wearing to [occasion]. Comment only on the clothes: proportions, colour, how formal it looks. Give me one change using something from my list. Don’t comment on my body.
That final sentence in Prompt 9 is worth keeping. A model will answer the question you ask, and you can choose what it looks at. Paste real forecast numbers into Prompts 2 and 8 too, rather than assuming it knows the weather where you are.
For the ideas behind Prompt 4, our guide to getting more outfits from fewer clothes goes further.

What ChatGPT can’t know
Even a perfect list leaves out a lot. It helps to know where to trust your own judgement instead.
| It doesn’t know | Why it matters |
|---|---|
| How a piece fits you today | A jacket can look right on paper and pull across the shoulders |
| How fabric feels | It can’t know that a jumper itches after an hour |
| What’s clean | Say so, or it will plan around the shirt in the laundry basket |
| What you wore last week | Unless you tell it, repeats are invisible |
| Your office, your friends, your mood | “Smart casual” means different things in different rooms |
| True colour | Photos shift with light, so shades next to each other may be misjudged |
If fit is the question, no prompt will replace a mirror. Our guide on how clothes should fit covers what to check. We’ve also written more generally about what AI can and can’t do for your wardrobe.
Try-on images are a preview
In October 2026, TechCrunch reported that ChatGPT had added a virtual try-on button to its shopping results, which uses your own photo to show how an item might look. Treat that as a picture of an idea. The report doesn’t address how accurately it shows fit or sizing, and a generated image doesn’t tell you how the item will hang on you.

Think before you upload photos
A photo of a shirt on a hanger says little about you. A mirror selfie shows your body and perhaps your home, so it’s worth knowing the settings.
- Training. OpenAI says it may use content from Free, Plus and Pro accounts to train its models. You can opt out by turning off Improve the model for everyone under Settings, then Data controls.
- Feedback. Even after opting out, pressing thumbs up or thumbs down can make that whole conversation available for training.
- Temporary chats. These don’t appear in your history, don’t update memories and aren’t used for training while they remain temporary.
Check these pages yourself before you rely on them, since settings change. If you’d like to see how a wardrobe app answers the same questions, we explain how we handle photos in Amalook.
When a chat stops being enough
A pasted list works well for occasional questions. It gets tiresome if you want to plan weeks ahead or track what you wear, since you’re maintaining the database yourself. That’s the gap we’re building Amalook for. It isn’t released yet, but the idea is that your wardrobe stays in one place, built from your photos, so each suggestion starts from what you own.
Common questions
Can ChatGPT plan outfits from photos of my clothes?
It can read photos you add to a chat, on free and paid plans, and suggest combinations. OpenAI says image descriptions can be wrong and counts only approximate, so check that it has identified each piece correctly before relying on it.
Will ChatGPT remember my wardrobe?
Not reliably. OpenAI says Memory doesn't keep every detail from every conversation. A written list that you paste into each new chat is the safer way to keep it working from what you own.
Are my outfit photos used to train ChatGPT?
On Free, Plus and Pro plans, OpenAI says it may use your content to train its models unless you turn off Improve the model for everyone in Settings under Data controls. Temporary chats aren't used for training while they remain temporary.
Sources
Checked on 8 November 2026. Listings change, so check them again before you decide.


