What platforms does Scento MCP support?

Scento MCP is available on Web App, MacOS, Windows, Linux and Browser extension.

What are the limitations of Scento MCP?

Coverage is limited to Scento's own catalogue of roughly 1,400 perfumes, so obscure houses and very recent releases may be absent or thin. The occasion list is fixed, and the docs tell callers to pick the nearest value, which can produce blunt advice for niche use cases. Data quality caveats are visible in the sample: some fields such as year and perfumer return null, so assistants must handle gaps.

Scento MCP

Name: Scento MCP
Author: Scento MCP

A free perfume knowledge server that gives AI assistants a nose

Intermediate API

Visit Scento MCP Back to AI tools

Own Scento MCP? Verify ownership to get a dofollow link

FeaturedHand-picked by aiwizard

AI Angels400+ customizable AI girlfriends for NSFW chat, voice calls, and roleplay

Candy AIExperience the future of companionship with AI friends who chat, role-play, and connect with you on a human level.

ourdream.aiA vast AI roleplay platform with millions of characters for chat, images, video and voice

Screenshot of Scento MCP, A free perfume knowledge server that gives AI assistants a nose

What is Scento MCP?

Scento MCP is a developer tool. Scento exposes its perfume database to AI assistants through a free MCP server and REST API. Twelve read only tools cover note pyramids, similar scents, side by side comparisons, raw materials, houses, occasion edits and a wardrobe builder, all with links back to scento.com and no API key or account required.

Scento's fragrance offering at this URL is a developer connector, not a shopping front end. The same perfume data that powers scento.com is exposed two ways: as a remote Model Context Protocol server at scento.com/mcp, and as a plain REST API under scento.com/api/perfume/v1. Neither asks for a key or an account. The dataset behind both covers more than 1,400 designer and niche perfumes, over 400 raw materials, and Scento's own fragrance family system. Every response is read only and carries a link back to a matching scento.com page, rendered in the user's language across ten languages. The practical value shows up when someone asks an assistant what a scent smells like. The server ships twelve tools that map to distinct perfume questions: searching by name, house, note or mood; pulling a full note pyramid, family, character and season for one perfume; finding near neighbours that share materials; contrasting two to four perfumes side by side; explaining a single raw material; listing perfumes built on given notes in a specific tier; profiling a house; walking the family tree; picking scents for a fixed occasion list; composing a small wardrobe; defining terms such as sillage or chypre; and searching Scento's journal. Each answer arrives as a folio with a quotable lede, structured detail, a citation line and a canonical page link. Setup is deliberately thin. Claude web and desktop accept the URL through the connectors menu, Claude Code takes a one line command, ChatGPT works through developer mode, and Cursor, VS Code and Windsurf read the same JSON block. Anyone building without MCP can hit the REST routes with curl instead. The honest framing: this is not a recommendation engine of its own. It hands curated fragrance knowledge to whatever assistant or app the user already runs, so the quality of the final advice depends on that host. Consumers wanting to browse can use the Scento perfume database directly, while the developer angle is clearly aimed at people building chat assistants, fragrance apps or shopping tools that need reliable notes, houses and comparisons rather than invented ones.

How do you use Scento MCP?

1Skim the REST and MCP documentation
Open the developer docs to see the two access paths and the twelve available tools before wiring anything up. The API and MCP pages list routes, parameters and example responses.
2Try a REST call with curl
Request a single perfume portrait from scento.com/api/perfume/v1/perfumes/layton and inspect the folio shape. No header or token is needed.
3Add the MCP server to Claude
In Claude, open Settings, then Connectors, add a custom connector and paste https://www.scento.com/mcp. The server is also listed in Claude's connectors directory.
4Configure an editor client
For Cursor, VS Code or Windsurf, add an mcpServers entry with type http and the scento.com/mcp URL. Claude Code uses the one line add command shown in the docs.
5Test real fragrance questions
Ask the assistant for a perfume portrait, a side by side contrast and a note explanation to confirm the tools route correctly. Check that returned scento.com links resolve in your language.
6Browse the underlying database
Compare what the API returns with the public perfume database pages to sanity check notes, families and house details before trusting them in an app.

Pros and cons

Pros

No API key or account: the MCP endpoint and REST routes answer immediately, which removes a common onboarding blockerAI
Each response is structured as a folio with a quotable lede, detail object, citation string and canonical page link, so citations are hard to fumbleAI
Twelve distinct tools separate concerns cleanly, so an assistant can search, portray, compare and recommend without overloading one vague endpointAI
Links and answers come back in ten languages, useful for multilingual assistants and storefrontsAI
Read only design keeps integration simple and avoids token management entirelyAI

Cons

Coverage is limited to Scento's own catalogue of roughly 1,400 perfumes, so obscure houses and very recent releases may be absent or thinAI
The occasion list is fixed, and the docs tell callers to pick the nearest value, which can produce blunt advice for niche use casesAI
Data quality caveats are visible in the sample: some fields such as year and perfumer return null, so assistants must handle gapsAI
Read only endpoints mean nothing can be written back, so no user collections, ratings or inventory trackingAI
No stated uptime commitment, rate limits or versioning policy on the page, which matters if an app depends on it in productionAI

How much does Scento MCP cost?

Pricing

Free. The page states the MCP server and REST API are free and read only, with no API key, account or sign-up needed. No paid tiers are mentioned.

Learn more

Integrations

Remote MCP server over streamable HTTP for Claude, Claude Code, ChatGPT developer mode, Cursor, VS Code, Windsurf and any MCP client, plus a parallel REST API for anything that can make an HTTP request.

Learn more

Features

Twelve read only tools: perfume search, perfume portrait with note pyramid and wear data, similar scent neighbours, side by side contrast, note and material portraits, matching perfumes by note and tier, house profiles, the Scento family map, occasion edits, a wardrobe composer, a perfumery glossary and journal search. Responses include a quotable lede, detail, citation and scento.com link in ten languages.

Learn more

Frequently asked questions about Scento MCP

How much does Scento MCP cost?
Free. The page states the MCP server and REST API are free and read only, with no API key, account or sign-up needed. No paid tiers are mentioned.
Does Scento MCP have an API?
Yes, Scento MCP offers an API.
How do you use Scento MCP?
The walkthrough on this page covers 6 steps: 1. Skim the REST and MCP documentation 2. Try a REST call with curl 3. Add the MCP server to Claude 4. Configure an editor client 5. Test real fragrance questions 6. Browse the underlying database.
What platforms does Scento MCP support?
Scento MCP is available on Web App, MacOS, Windows, Linux and Browser extension.
What does Scento MCP integrate with?
Remote MCP server over streamable HTTP for Claude, Claude Code, ChatGPT developer mode, Cursor, VS Code, Windsurf and any MCP client, plus a parallel REST API for anything that can make an HTTP request.
What are the limitations of Scento MCP?
Coverage is limited to Scento's own catalogue of roughly 1,400 perfumes, so obscure houses and very recent releases may be absent or thin. The occasion list is fixed, and the docs tell callers to pick the nearest value, which can produce blunt advice for niche use cases. Data quality caveats are visible in the sample: some fields such as year and perfumer return null, so assistants must handle gaps.