Garden Knowledge Base

The articles are not the database.

The knowledge base is the normalized layer that should eventually power articles, maps, calendars, diagrams, crop recommendations, garden plans, and diagnostic tools.

Data inventory

The current normalized database is generated from the Rare Seeds fact import plus initial project-authored model datasets. It is not complete, but it gives veggie.farm a data-first spine.

Architecture layers

Plant coverage

The plant database is normalized from the larger Rare Seeds import into records with taxonomy, propagation, planting, climate, site, morphology, ecology, harvest, safety, media, and provenance sections.

Sample normalized plants

Source inventory

The source registry is now a first-class dataset. Each source has organization, official URL, licensing notes, API status, update frequency, geographic coverage, reliability score, recommended use, citation format, and cache strategy.

Decision models

Decision rules describe what the future app needs to ask and which data layers are required. The answer should be computed from structured records, graph relationships, local climate, soil, and user goals.

Image catalog

Images are metadata records, not a loose folder dump. Local generated illustrations and external source-image references both receive catalog entries, license notes, source links, keywords, and copy status.

Build workflow

Regenerate the normalized knowledge base and graph with:

npm run build:kb

That runs:

node scripts/build-knowledge-base.mjs
node scripts/build-knowledge-graph.mjs

Field notes

The most important shift is architectural: source data becomes normalized JSON, normalized JSON feeds the graph, the graph feeds decision engines, and the Observable site becomes the view layer.