Skip to content
planned claude-skill · google-sheet

/catalog-search

Translate fuzzy designer language into a ranked FF&E spec list.

sourcing search specs
Inputs
  • natural-language query
  • structured filters (category / price range / manufacturer / certification / color / material / lead time)
Outputs
  • ranked product list (name
  • manufacturer
  • SKU
  • price
  • lead time
  • certifications
  • image URL
  • source)
Distribution
  • claude-skill
  • google-sheet

Catalog Search is the front door to the Canoa FF&E dataset. A designer drops a natural-language query into Claude (or a Sheet template) and gets a ranked, structured list of real products that match — not a hallucinated list, not a generic catalog scroll.

Why this exists

Conversational catalog search is the killer use case for AI over a curated FF&E dataset — translating designer-shaped language (“a warm-toned lounge chair under $1k that ships fast, GREENGUARD if possible”) into structured filters, then returning real products. It’s what the original Canoa designer assistant proved was the highest-value AI surface. This skill ports that pattern out of the dead canvas and into the tools designers already work in.

Typical queries

  • “Three task chairs around $600, fabric, mid-century-adjacent, in stock”
  • “Coffee tables for a clinic waiting room, recyclable, under $400”
  • “What’s the cheapest GREENGUARD-certified lounge chair we have?”

Distribution

  • Claude skill — interactive: ask, refine, re-ask in the same thread.
  • Sheet template — drop the dataset into a tab; formula columns to filter and rank for one-off lookups.

Status

Planned — pending dataset export from the Canoa Postgres + a one-time Parquet/SQLite snapshot. Source materials in ~/Documents/Canoa/dataset-extraction-plan.md.

Install

/catalog-search is part of the Canoa agent. Copy the install prompt below and paste it into Claude — it walks you through install, sets up the master schedule Sheet, and brings the rest of the toolkit with it.

Related skills