Fashion MCP Directory
The open directory of fashion MCP (Model Context Protocol) servers. Every store is probed for what an AI shopping agent can actually answer, scored, and scanned for security. Find your brand, see where you sit against your peers, and see what changes when your catalog is enhanced.
The fashion MCP server list, re-verified daily. 1 of the 330 stores listed have already moved past a native endpoint. Filter by capability, protocol, and readiness score to see who.
Loading directory...
The score is the endpoint an agent reaches today. The smaller number is where that catalog started.
Every listing is measured on the endpoint the store currently serves. Where a retailer has moved to Enhanced, the headline is that endpoint's score, shown in orange, with the native score beneath it for reference. Scores and scan status are re-checked on a published schedule, and listings are never paid.
Same protocol, same tool names. What changes is the data behind them and the discovery services wired in front. Every endpoint in the directory is free for builders to query - the tier describes what the retailer runs, not what you pay.
Tier 1 · Native
Whatever the catalog exposes today - usually product titles, prices, and stock over keyword search. Listed, scored, and scanned at no cost.
Tier 2 · Enhanced
We tag your catalog with 10–15 fashion attributes per product, then wire the full discovery stack in front of it: hybrid search, complete the look, you-may-also-like, collaborative filtering, and brand recommendations, all configured as MCP servers behind the same tool names.
Every MCP server speaks the same four tools. You don't need to learn how to build an MCP server to start. Three steps to connect, and you are building against real catalogs, live prices, and live stock.
Step 1
Filter by category, region, protocol, and readiness score - or add the Directory MCP and let the agent choose per query.
Step 2
Copy-paste config for Claude, ChatGPT, and any MCP-speaking agent sits on every listing.
Step 3
Real catalogs, live prices, live stock - and an outfit at the end of it.
Step 4
Seven and more things builders and merchandisers are putting together on top of these endpoints right now.
Each one is a real sequence of tool calls against endpoints already in the directory.
For Builders
Can't pick a favorite? Shoe Tournament turns shopping indecision into a game. You judge every head-to-head matchup until your winning pair is crowned, ready to shop in one tap.
Prompt: I want to build a "shoe world cup" web app — a bracket-style tournament for finding the shoes I want. Product info can be fetched from mcp. Given my criteria, it should pick a pool of matching shoe candidates, then show them two at a time so I can pick until a final winner remains. At the end, show the full tournament bracket laid out from top to bottom.
For Creators
Your Instagram feed, built straight from the catalog. Pick a retailer, describe a category and mood, then choose from live product candidates, design right on the images Canva-style, and export ready-to-post PNGs.
Prompt: Build an "IG feed builder" web app for creating Instagram content. Product info can be fetched from MCP. Given a retailer and my category/mood, it should show matching product candidates, let me pick the ones I like, design on the images with a Canva-like editor, and export the finished posts as PNGs.
For builders
An agent that dresses a shopper head to toe from real inventory across every store you connect.
search_products(occasion="casual date", seasonality="summer", category="dress", max_price=250) complete_the_look(product_id="drs-4471")
→ 13 dresses in stock, ranked by occasion fit → outfit: sandals $88 · tote bag $64 · earrings $32 → all three ship to US in 3 days
For retailers
The agent reads what shoppers searched for last week, clusters it by attribute, and assembles a landing collection from in-stock SKUs.
cluster_queries(window="7d", min_volume=50) build_collection(cluster="the fall cardigan edit", in_stock=true, max_price=300)
→ 6 clusters above threshold → collection: 28 SKUs, all in stock → ranked by attribute match, not bestseller order
For builders
Vague, human briefs - a budget and a vibe - resolved into products that are actually in stock in the right size.
For retailers
Merchandising view: which shopper intents your catalog can answer, and which ones return nothing.
For builders
A shopper connects order history from the stores they buy from and the agent builds a closet out of real SKUs, already tagged.
For builders
A standing agent that polls endpoints for a described item and pings when it appears or comes back in stock in the right size.
For retailers
Next season's gaps stated as attributes with volume behind them, ready to hand to a buyer or a supplier.
FOR FASHION BRANDS/RETAILERS
Agentic commerce is coming to fashion whether your catalog is ready or not. Get enhanced, get listed. We build and run the endpoint end to end - your catalog, your branding, your SLA.
What the meeting covers
Run a free scan. We probe your public endpoint or storefront and email you the readiness score, the failing queries, and what enrichment would change.
Step 1 of 2 - no meeting, no sales call.
For builders
An MCP (Model Context Protocol) server is a small API layer that lets AI agents talk to a store's catalog through a shared, open standard. A fashion MCP server exposes tools like product search, similar styles, outfit suggestions, and review Q&A, so agents can query live products, prices, and stock without scraping the storefront. With a single MCP server, a retailer makes its products accessible to every AI shopping agent, and anyone, whether a shopper, a merchandiser, or a new store owner, can quickly build AI-powered shopping experiences on top of the catalog.
The protocol is open and free, and many stores run their own MCP server at no cost beyond hosting. In this directory, listing and scoring any server is always free. Enhanced servers, where YesPlz tags the catalog and runs the full discovery stack, are a paid production service.
We probe each server with a fixed set of real shopper queries across the four standard tools and score what an agent can actually answer: attribute depth, occasion queries, result relevance, and price and stock accuracy. Scores are re-verified daily and the methodology is public. Listings and scores are never paid, ever.
Two paths. Build it yourself: expose your catalog over MCP with search at minimum, and the open guides in our Build section show how to test it against the directory's scoring. Or skip the build: YesPlz runs an Enhanced MCP server for you, with your catalog tagged, the full discovery stack wired in, and your branding on your domain. Either way, listing and scoring in the directory is free.
Every listing includes copy-paste config for Claude, ChatGPT, and any MCP-speaking agent. Add one store, or add the Directory MCP itself and your agent picks stores per query with search_retailers and suggest_retailers_for_intent.
Free, always. If your store has a public MCP server, request a scan and we'll probe, score, and list it. No server yet? Run the free scan on your storefront and we'll email you the readiness score and what a server would unlock. Listings are never paid and never ranked by payment.
MCP (Model Context Protocol) is an open standard created by Anthropic and now supported across the industry, including Claude and ChatGPT. It lets AI agents read and query a store's catalog: search, compare, style. That's the discovery side. UCP (Universal Commerce Protocol) is a commerce standard led by Google with partners including Shopify, and it covers the transaction side: carts, checkout, payment. They're complementary. An agent can find the dress over MCP and buy it over UCP.
Not at this point. We're starting with MCP, then adding UCP and other channels. Stay tuned.
Two ways: some are brands we admire and added ourselves, others asked to be listed. Either way, we verify MCP availability and security before anything goes live. Want your brand on the list? It's free — just reach out.
