Ask a general-purpose assistant what a perfume smells like and you usually get a confident paragraph that mixes up flankers, invents notes, or recommends something that was discontinued years ago. Fragrance is a niche where structured, curated data makes a big difference.
So we put the 10ML Fragrance Index behind a public MCP server. It is free, read-only and needs no account or API key. You add one URL and your assistant can answer perfume questions with real note pyramids, accords, families and editor-written context.
https://mcp.10ml.com/mcp
Transport is Streamable HTTP, authentication is none.
The index is curated by the fragrance editors at 10ML, a European perfume decant shop:
Claude (web and desktop): Settings, Connectors, Add custom connector, paste https://mcp.10ml.com/mcp. It is also listed in the Claude connectors directory as "10ML Perfume Finder".
Claude Code:
claude mcp add --transport http perfume-finder https://mcp.10ml.com/mcp
Cursor (~/.cursor/mcp.json):
{
"mcpServers": {
"perfume-finder": { "url": "https://mcp.10ml.com/mcp" }
}
}
VS Code (.vscode/mcp.json):
{
"servers": {
"perfume-finder": { "type": "http", "url": "https://mcp.10ml.com/mcp" }
}
}
Clients that only speak stdio can use the thin wrappers, which forward to the same endpoint:
npx -y @10ml/perfume-finder-mcp
uvx perfume-finder-mcp
Behind those questions the assistant picks from 16 read-only tools: search, full perfume profiles, side-by-side comparison, similar perfumes (with a "fresher / sweeter / darker" steer), editorial alternatives, notes, houses, a perfume glossary, layering, occasion picks, gift ideas and guide search. Every tool takes a locale (en, de, fr, es, it, ro, el, hu, cs, pl) and returns structured content plus a short text summary.
The same answers are available over plain HTTP, described by an OpenAPI 3.1 file:
curl "https://mcp.10ml.com/api/v1/notes/bergamot"
curl "https://mcp.10ml.com/api/v1/similar?perfume=Baccarat+Rouge+540&steer=fresher"
curl "https://mcp.10ml.com/api/v1/houses/amouage"
A note lookup returns what the note smells like, its family and mood, what it pairs with, similar notes and famous perfumes that feature it, so it drops straight into a tool call, a chatbot or a recommendation widget.
com.10ml/perfume-finder
Questions or ideas for new tools are welcome in the GitHub issues.
10ML Team