Fragrance MCP

The perfume MCP server built for AI coding agents

If you searched for perfume MCP, fragrance MCP server, or Model Context Protocol perfume data, you are in the right place. scentrev-mcp exposes 14 tools so LLM agents can recommend, compare, and explain fragrances with real community metrics — not hallucinated note lists.

Why agents need a dedicated fragrance MCP

General web search gives inconsistent perfume descriptions. A fragrance MCP returns normalized JSON: star ratings on a 0–5 scale, wear performance buckets, season scores, accords with percentages, and “reminds me of” links backed by thousands of votes. That is what makes automated perfume assistants trustworthy.

  • Structured wear metrics

    Longevity, sillage, projection, season fit, and time-of-day scores with confidence categories — not scraped marketing copy.

  • Discovery & blending

    Filter by notes and accords, sort by rating or engagement, blend note pyramids from multiple reference fragrances, or compare side-by-side.

  • Agent-native MCP

    Install in Cursor or Claude Desktop with a Bearer token. Tools return ratio-safe JSON capped at 10 rows so agents cannot mine the full catalog.

  • REST fallback

    Every MCP tool maps to a PostgREST RPC on the same backend if you prefer HTTP from your own stack.

14 MCP tools included

Search the catalog, apply advanced filters, fetch full profiles, list accords, find similar scents, compare fragrances, browse brands, and blend note pyramids from reference perfumes.

  • search_fragrancesFree-text search across the catalog
  • search_fragrances_filteredFilter by brand, season, notes, discovery sort
  • search_by_referencesBlend note pyramids from 2+ reference scents
  • list_brandsBrowse brands with fragrance counts
  • resolve_slugResolve informal names to slugs
  • compare_fragrancesSide-by-side wear and value metrics
  • get_fragrance_profileEvery metric for one fragrance
  • get_wear_summaryWhen and where it shines
  • get_identityGender lean and character
  • get_performanceLongevity and sillage scores
  • get_appreciationCommunity sentiment and love score
  • get_reminds_ofClosest-smelling alternatives
  • get_note_pyramidTop, heart, and base notes
  • get_accordsDominant accords, weighted

Connect in under two minutes

  1. Sign in and create a free API key (frag_live_…).
  2. Add the MCP server URL https://api.scentrev.com/mcp/ with Authorization: Bearer … in Cursor or Claude Desktop.
  3. Read the documentation for tool examples and filter cheat-sheets.

Open MCP documentation

Related pages