To add your Wearly Labs AI profile to Claude, create a Project, put the .md file in the project's files and paste a standing instruction into the project's instructions. Every chat you start inside it then answers against your palette and your cuts, with nothing to attach again.
On this page
- Before you start
- Step 1: open Projects
- Step 2: name it and say what it is for
- Step 3: add the file
- Step 4: add the instructions
- Step 5: test it before you trust it
- What a correct answer looks like
- What to ask once it works
- Keeping it current
- If you do not want a project
- Troubleshooting
- The AI profile: common questions
- Related reading
This works on every Claude plan, the free one included. As of October 2026, Anthropic's help center says Projects are available to all users, with a limit of five projects on a free account. A Project holds both halves of what this needs, the file and the standing instruction for what to do with it, so you build it once and every conversation inside it starts already knowing your palette.
Your analysis comes with an AI profile: a plain-text Markdown file with your color season, a verdict and a reason for all 50 colors, your body shape and line, the cuts to wear and the cuts to avoid, and a short set of rules for the assistant that reads it. It holds no photos. More on the file is on the Connect to AI page.
Before you start
Get the file from your analysis page, under Connect to AI. Press Download .md. It is named after your season, for example wearly-deep-autumn.md. The screenshots below show our test copy, saved as wearly-labs-profile.md.
Check the name when it arrives. If your browser added .txt, you end up with wearly-deep-autumn.md.txt. Claude will still read it, because Anthropic's list of supported file types includes TXT, but rename it back so you can find it later.
Step 1: open Projects
A Project is Claude's workspace for one topic: it keeps its own chats, its own files and its own instructions. On a phone, open the menu and pick Projects. On the web, go to claude.ai/projects. A fresh account has none yet, so New project is the button you want.
Step 2: name it and say what it is for
Claude asks two questions when you create a project.
What are you working on? Put the name here. "Wearly Labs" is what we use in the screenshots; "My style" works just as well.
What are you trying to achieve? "Help with clothing style" is enough.
Both fields are for you. Anthropic's help center notes that Claude does not have access to a project's name and description, so neither one changes an answer. The instruction that does comes in step 4.
Step 3: add the file
The project opens with two buttons under its description: Add files and Add instructions.
Press Add files. A sheet slides up headed Project Files, empty, saying "No files yet. Add files for Claude to use in Wearly Labs." Press Add Files at the bottom and pick your profile from wherever your phone put it.
It should come back listed with a size next to it. That sheet is the whole trick: everything in it is available to every conversation in the project, without you attaching anything again. On the web the same place is the project knowledge on the right of the project page, and Anthropic says anything uploaded there is used across all chats in that project.
Step 4: add the instructions
The file tells Claude who you are, and it opens with a few rules for the assistant that reads it. The project instructions add what a file does not: a standing role and the shape of every answer. Claude uses them for all the chats within the project.
Press Add instructions and paste this in. On the web the button is Set project instructions, followed by Save instructions.
My personal style profile is in this project's files. Act as my shopping and styling assistant. Before answering any question about what to buy or wear, check it against my color palette, my body shape and line, and my recommended cuts. Give me the verdict first (Wear, Depends or Skip) and the reason second, in one or two sentences. If a color is a Depends, say where it can go. If something does not work, name a specific alternative that does. Never recommend anything the file lists as not my colors or as a cut to avoid.
Be specific about the shape of the answer you want. Anthropic's prompting guide says Claude responds well to clear, explicit instructions, and that giving the reason behind an instruction helps. If you would rather have a short paragraph than a blunt verdict, say so here, with the reason ("I read these on my phone in a shop"), rather than re-asking in every conversation.
Step 5: test it before you trust it
Start a chat inside the project and ask something you can already check:
What are my three best colors, and what do I wear instead of black?
You should get names from your own palette and your profile's own substitute for black. If the answer is about black being slimming, the file is not being read. Check that it is in the project's files and not attached to a single chat, then start a fresh chat inside the project.
Then ask about one piece from your own Skip list, in the same plain words:
What do you think about skinny jeans for me?
The answer should be a Skip with the reason your page gives for it. Swap in a piece from your list if skinny jeans are not on it.
What a correct answer looks like
The sample analysis is the real result of co-founder Filip Poláček: Deep Autumn, with a Trapezoid shape and a medium line. With his profile in a project, the answers for these six pieces should match his page.
| Ask Claude about | Verdict | The reason in the profile |
|---|---|---|
| A rust t-shirt | Wear | Warm, dark and softened, just like his own natural coloring. |
| Black trousers | Depends | Black stays below the waist: clear and deep, but its coolness sits wrong close to his skin. |
| An orange belt | Depends | Vivid enough to spark an outfit, too loud spread over one. |
| A fuchsia scarf | Skip | Its icy pink base clashes with the warmth under his skin. |
| Straight denim | Wear | Its waistband and seat are cut for this frame, so the leg falls straight without pulling. |
| Skinny jeans | Skip | Tighter than his frame wants, so the straight leg clings instead of skimming. |
For a color, Depends means yours in one place only: below the waist, or as an accent. Skip means not yours anywhere, not even as an accent. So the first test has a right answer too: in his profile espresso is the black, and black itself is a Depends for trousers and shoes. An answer that says "black works for everyone" has not read the file.
What to ask once it works
Claude reads photos you attach (JPEG, PNG, GIF and WebP), which changes which questions are worth asking.
- The rack. Photograph a rail in a shop. "Anything here in my palette?"
- The link check. Paste a product link. "Verdict, and why." Claude opens the page when Web search is switched on in the + menu of the chat box. If it cannot, paste the product title and description instead.
- The wardrobe audit. Photograph a shelf. "What earns its place, what does not."
- The pack. "Four days in Lisbon in June. Pack me from my palette, nothing I would have to iron."
- The second opinion. A changing room mirror photo. "Is this cut doing what my analysis says it should?"
When you cannot decide in a shop, the sample analysis (a Deep Autumn) gives this test: hold the piece against your skin in daylight. If it still looks rich and warm rather than washed out, take it home. Take that photo, send it to Claude and ask whether it agrees with your profile.
Keeping it current
The profile is plain text, so edit it. Add sizes, add a brand you buy often, add the thing you will never wear regardless of what the analysis says. Keep the headings: they are what the assistant reads to tell your colors from your cuts. The same goes in the other direction. If you would rather not hand a line to an AI assistant, such as the height and weight under Body measurements, delete that line before you upload.
Then replace the file in the project rather than adding a second copy: two versions of your profile in one project is how you get answers that contradict each other.
If you do not want a project
Attach the file at the start of a conversation and paste the instructions from step 4 as your first message. Anthropic says files can go to an individual chat or to a project's Files section, which persists across conversations, so you do the setup again next time. The answers come from the same file.
On a phone the quicker move is Copy on the Connect to AI screen: paste the whole profile as your first message with the instructions under it, and skip file handling entirely.
| Way in | Where it lives | The next chat | Use it for |
|---|---|---|---|
| Project files | The project's Files section | Already there | Daily use |
| Attach to a chat | The + menu in the chat box, Add files or photos | Not there, attach it again | One question |
| Paste the text | Your first message, from Copy | Not there, paste it again | A phone, or no file handling |
Troubleshooting
Claude answers generically. Check the instructions saved. The file carries a few rules of its own, but only the project instructions give Claude a standing role and an answer shape in every chat. Then start a fresh chat inside the project.
It says it cannot find the file. Make sure you added it to the project's files, not to one chat, and start a fresh chat in the project.
It hedges. Add "Give one verdict and one reason." to the instructions.
Long conversations drift. Say "check my profile again" and it re-reads the file. Or start a new chat, which also uses less of your usage limit: Anthropic lists the current conversation length among the factors.
An answer mentions something that is not in your file. That can be Claude's memory. It is on by default for Free, Pro and Max, and each project keeps its own, separate from your other chats. To test the file alone, turn Memory off for one chat in the + menu before your first message.
The AI profile: common questions
No. As of October 2026, Projects are available on every Claude plan, free included, and a free account can create five of them. Claude's pricing page lists Projects as "Up to 5" on Free and "Yes" on Pro and Max. If you are at your five, or your app does not show Projects, use the per-conversation fallback above: the answers come from the same file, the setup just does not persist.
Yes. A Markdown file is plain text with a different extension. Anthropic's list of supported file types names TXT and does not name MD, so if an app ever refuses the file, rename it to .txt and add it again. Nothing inside changes.
In a project that holds only your profile, yes. Anthropic says a project switches to retrieval, where Claude searches the files instead of loading them, only when its knowledge approaches the context window limit, and only on paid plans. A profile is plain text in the tens of kilobytes, small next to any context window. If you keep large documents in the same project, name the file in your question, which Anthropic suggests for projects that use retrieval.
You can, but do not. Two profiles in one project leave Claude unable to tell which one you mean. Make a second project for a second person, which is also how a household with two analyses should set this up. On a free account that second project is one of your five.
Yes. Anthropic's Claude Academy says conversations, projects, memory and uploads work the same on the web, your phone and the desktop app, so the project you build on a laptop answers on a phone.
No. The file is in your account, and we have no visibility into what you do with it. That is the point of shipping a file rather than a chatbot. The file is text only, and what Claude does with a chat falls under Anthropic's own privacy terms and your Claude settings.
