livostyle-catalog-mcp

MCP server giving any AI agent (Claude Desktop · Cursor · Cline · Cowork) live access to 2,766+ women's fashion products from Livostyle.com. Drop-in shopping context for shopping agents, outfit recommenders, fashion chatbots.

Install

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "livostyle-catalog": {
      "command": "npx",
      "args": ["-y", "livostyle-catalog-mcp"]
    }
  }
}

Restart Claude. The Livostyle tools appear in the tool list — your Claude can now recommend outfits and search a real fashion catalog.

Cursor / Cline / Cowork

Same JSON config — drop into the MCP-server section of your client's settings.

What it does

Six tools that any LLM agent can call:

Try the Stylist — no install needed

The same MCP-backed brand context, running in Claude.ai as a public shareable conversation:

• 🌸 Garden wedding guest outfit ($150 budget, June, Albuquerque) → view chat

• 🏖️ 5-day Cabo beach vacation capsule ($250 budget) → view chat

• 🎁 Birthday gift for a 24yo NYC quiet-luxury sister ($80 budget) → view chat

• 💬 Additional stylist demo → view chat

Each link opens a full conversation showing live product recommendations, prices, and links pulled from Livostyle’s catalog. Continue any of them with your own Claude account — the project knowledge transfers.

Examples (what an agent can do)

User: "Find me a floral midi dress under $50 with at least 5 reviews"
→ Agent calls search_products(query: "floral midi", max_price_usd: 50, min_rating: 4.5)
→ Returns 3 ranked picks with prices, ratings, links

User: "What should I wear to a garden wedding in summer? Budget $100."
→ Agent calls recommend_outfit(occasion: "garden wedding guest", budget_usd: 100, season: "summer")
→ Returns 3 complete outfit suggestions with reasoning

User: "Show me everything in the Coachella collection"
→ Agent calls get_collection(handle: "coachella-outfits")
→ Returns 475+ festival-ready picks

Data source

Live catalog from Livostyle.com, mirrored weekly via GitHub Actions to:

• 🐙 GitHub: arturayupov/womens-fashion-catalog-open-data

• 🤗 Hugging Face: arturayupov/womens-fashion-catalog

• 📦 NPM (this package): livostyle-catalog-mcp

The MCP server fetches products.json from the GitHub mirror on first use, caches in memory, refreshes every 6 hours.

What's in the catalog

• 2,766+ active women's fashion products

• 158 curated collections

• 4.76 avg rating, 15,937 total reviews

• 99% products with 5+ reviews · 99% with 4+ images

• Price range: $18–$120 USD

• US shipping, free over $97, 7-day returns

• Categories: Dresses, Tops, Bottoms, Outerwear, Swimwear, Activewear, Two-Piece Sets, Accessories, Shoes, Jewelry, …

Related

• 🎬 Livostyle Lookbook — 50 short 9:16 outfit videos generated from this catalog. MIT, $0 cost, ready for YouTube Shorts / TikTok / Reels.

• 📊 Fashion-Agent-Benchmark — open eval framework for AI shopping agents. 200 tasks × 5 evaluators using this catalog as ground truth.

License

MIT — see LICENSE. Free for commercial use, AI training, research. Attribution appreciated.

• 🎀 Free interactive tool — Wedding Guest Outfit Finder — 5-question quiz, 4 curated picks. Vanilla HTML, MIT, fork-able.

Why this exists

We're in the AI shopping era. ChatGPT, Claude, Perplexity, Gemini are becoming the new search bar. Stores that publish their catalogs as MCP servers get drop-in distribution to every agentic shopping client. This is our experiment in being there first.

If you build something with it, drop us a line at info@arcada.store — we may feature your project.

Contact

• Email: info@arcada.store

• Site: https://livostyle.com

• Issues: GitHub Issues