Model Context Protocol
Ask your AI assistant what to wear — and have it answer from your own closet
Clad9 speaks the Model Context Protocol, so Claude, ChatGPT, Gemini or anything you've built yourself can reach the wardrobe you've already catalogued. Not a chatbot's guess about clothes in general — the trousers in your drawer, the shirt in the wash, and the dinner on Thursday.
One endpoint: https://clad9.com/mcp
What your assistant can do
Seventeen tools, named for the things people actually ask. Every one of them goes through the same code the Clad9 app itself uses, so an outfit asked for in a chat window is the same outfit, on the same allowance, as one asked for on the website.
Read your wardrobe
- "What navy do I own?" — natural-language search over your real closet.
- "What's in the wash?" — the laundry basket, and how full it is.
- "What have I worn this week?" — past looks and what you actually wore.
- "What's on Thursday?" — your upcoming events and what Clad9 makes of them.
Get dressed
- "Something for a client dinner" — three looks, with the picture.
- "Same look, different shoes" — one piece changed, the rest kept.
- "Dress me for Thursday's dinner" — the calendar event, specifically.
- "I wore that" — marks it worn, and sends the washable pieces to the basket.
Outfit pictures come back as images, so a client that renders them shows you the look rather than describing it.
Connecting one
You need a Clad9 account with a wardrobe in it. Everything below is set up from Connections → AI assistants inside Clad9, and every connection is separate — connect your laptop and not your phone, allow one assistant more than another, and disconnect either without touching the rest.
Claude
Add https://clad9.com/mcp as a custom connector.
Claude finds its own way in — you'll land on a Clad9 screen that names Claude and
lists, in plain words, everything it is asking for. Approve it and you're done;
there's no key to copy.
In Claude Code, from a terminal:
claude mcp add --transport http clad9 https://clad9.com/mcp
ChatGPT
Add https://clad9.com/mcp as a connector in
settings. Same approval screen, same result. Clad9 also ships rendered cards, so a
recommended outfit arrives as something you can look at and tap rather than a list
of garment names.
Gemini CLI, your own scripts, anything else
Create a connection key in Clad9, tick what it may do, and send it as a bearer token. That is the whole protocol surface — one HTTP endpoint, one header.
Authorization: Bearer <your connection key>
The key is shown once, when you create it. Name each one for the machine it lives on, so you can disconnect just that one later.
What it will never do
These aren't permissions you can grant and choose not to. They aren't offered at all, and that's deliberate: an assistant's context can hold a web page, an email or a calendar invite written by somebody who is not you.
- No photographs of you, ever. Your FaceBooth photos, the self-photos you upload and your fit-check pictures are never sent to a connected assistant, at any permission level, in any form. Clad9 will tell it which garments a fit check matched; it will not give it the picture. Photos of your clothes and the outfit images Clad9 draws can be shared — that's what makes the answer useful. A photograph of you is a different thing.
- Nothing destructive. No deleting garments, no account or plan changes, no payment details, no disconnecting your calendar. Everything an assistant can change, you can change back in the app in one tap.
- No adding to your closet. Cataloguing a garment is a photo you took, deliberately, with a duplicate check and a cost behind it. An assistant looping over that is a bill and a polluted wardrobe, so capture stays something a person does.
- No extra allowance. MCP doesn't sell anything. An outfit asked for here costs the same outfit it costs on the website, out of the same daily allowance.
What we keep, and what we can't
- A log you can read. Every request an assistant makes is recorded and shown back to you in plain language — "Asked for an outfit for 'client dinner'", "Sent 3 items to the laundry basket" — under the connection that made it. A log you can't read isn't consent.
- The assistant is not Clad9. Your question and what Clad9 sends back are handled by whoever makes that assistant, under their privacy policy and their retention and training terms, not ours. We can't delete it for you. Read their policy before you connect them.
- Disconnecting works immediately. We revoke the credential on our side, so it stops being able to read anything whether or not it still holds one.
The full detail is in our privacy policy, under "Connected AI assistants".
Catalogue your wardrobe first
An assistant can only answer from clothes Clad9 knows about. Photograph your closet, let Clad9 identify what's in it, and then point whatever you're already working in at it.