What is a perfume MCP server?

MCP in plain words

MCP, the Model Context Protocol, is an open protocol that lets an assistant call outside tools and read outside data. A chat app or code editor that supports it is an MCP client; the service it connects to is an MCP server, reached at one web address.

Without a tool, an assistant answers perfume questions from memory, and memory is where notes, launch years and houses go wrong. With a perfume MCP server it calls a tool, gets the facts from a maintained index and writes the answer from them.

What a perfume MCP server does

1. You ask in plain words: “What notes are in Dior Sauvage?” or “Something like Creed Aventus, but fresher.”
2. The assistant picks a tool from the server’s list, here get_perfume or find_similar_perfumes, and fills in its inputs.
3. The server answers with structured data from the 10ML Fragrance Index: note pyramid, accords, family, similar perfumes.
4. The assistant reads the result and replies in your language, with the facts and links it was given.

Here is one exchange as a raw call. Asking what bergamot smells like is one call to the explain_note tool; the first block is what the assistant sends, the second what the server returns (trimmed, called on 21 October 2026).

curl -X POST https://mcp.10ml.com/mcp \
  -H "content-type: application/json" \
  -H "accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"explain_note","arguments":{"note":"bergamot"}}}'

{
  "found": true,
  "note": {
    "name": "bergamot",
    "family": "citrus",
    "smells_like": "Sparkling, slightly floral citrus with a green, tea-like bitterness.",
    "mood": "bright, polished, uplifting",
    "pairs_with": ["lavender", "vetiver", "neroli", "black pepper", "musk", "cedar"],
    "similar_notes": ["lemon", "orange", "bitter orange", "mandarin", "tangerine"]
  },
  "famous_perfumes": [
    {
      "name": "Aventus",
      "house": "Creed",
      "family": "Chypre fruity"
    },
    {
      "name": "Silver Mountain Water",
      "house": "Creed",
      "family": "Fresh green"
    }
  ],
  "source": "10ML Fragrance Index"
}

The 15 tools cover search, one perfume’s profile, comparisons, similar scents, alternatives from other houses, taste-based picks, notes, houses, vocabulary, layering, occasions, gifts and 10ML guides. The perfume MCP server page lists each with an example question.

Fragrance API or perfume MCP server?

Both read the same index, so the facts match. Pick the API when your own code shows the data and the MCP server when an assistant should find it itself. The perfume API page has the endpoints, code in curl, JavaScript and Python, and real responses.

What to look for in a perfume MCP server

• No key or sign-in, so the connector works the moment you add the URL.
• Clear tool descriptions: each tool says what it is for and what it is not for, so the assistant picks the right one.
• Structured output: named fields for notes, accords and family rather than one block of text.
• Few calls and a fast answer: a good answer in one call keeps the conversation quick.
• An honest size: a bigger index covers more rare perfumes, so check how many perfumes a server claims and how it does on the questions you care about.
• Your language: note names and guides in the language you ask in.

We put the free perfume MCP servers that need no key through 40 questions on 20 October 2026: 10ML Perfume Finder answered 39 correctly, always in one call, at a median of 232 ms. Larger indexes lead on breadth. The benchmark page has every score, the method and the data.

Use one in two minutes

1. Take the address https://mcp.10ml.com/mcp.
2. Add it as a connector: steps for Claude, steps for ChatGPT, or the perfume MCP server page for Cursor, VS Code and the rest.
3. Ask a perfume question in a new chat and check that the assistant names 10ML Perfume Finder among the tools it used.

Keep reading

• MCP server αρωμάτων
• API αρωμάτων
• Σύγκριση MCP server αρωμάτων
• Claude perfume connector
• Perfume ChatGPT app
• Perfume recommendation API

Questions