09.04

Data schema registry

Shared datasets have one canonical shape so tools interoperate.

Updated: Oct 3, 2026

On this page

Purpose

Shared datasets have one canonical shape so tools interoperate.

The standard

Canonical schemas live in [TEAM: org]/schemas as JSON Schema (for validation) plus a Markdown page here per schema.
Each schema has: name, version (semver), owner, fields (name, type, unit, required, description), example record, producers, consumers.
Breaking change = major version; producers and consumers are updated via issues tagged schema:<name>.

Registry

Schema	Version	Owner	Produced by	Consumed by
`project` — id, code, name, client (anonymized id), phase, units	1.0	`[TEAM]`	intake	all
`facade_panel` — panel id, system type, width_mm, height_mm, glass spec, frame depth_mm, location (elev/grid), weight_kg	1.0		Rhino generators (10.06), Revit export (11.06)	takeoff dashboards (09.08)
`curtain_wall_system` — manufacturer, series, type (captured/SSG/unitized/ultra-thermal), profile depth, thermal data	1.0		12.01 library	Rhino generators
`stair` — type, rise, run, flights, width, code basis	1.0		stair generator
`commute_od` — origin, destination, count, mode, year (census)	1.0		data pipeline	Omaha expressway dashboard
`[TEAM: add]`

Procedure — new schema

Draft JSON Schema + example → PR to schemas repo → add row here and a subpage.

Owner: [TEAM] · Last reviewed: 2026-09