Skip to content

Project

Facts about the whole project, under project in oryxkb/project.yaml. Every key is optional. For a data vendor, one project is usually one data product.

Example

project:
  name: marketdata-example
  description: Synthetic rents for the docs.
  owner: data-team
  license: CC-BY-4.0
  org: example-org
  team: research
  sources:
    vendor_a: Quarterly rent file from vendor A
    vendor_b: Annual vacancy survey from vendor B
  aliases:
  - name: market-data
    renamed: 2026-11-02
  retired_names:
  - name: old-rents
    successor: marketdata-example
    reason: merged into marketdata-example
  removed_datasets:
  - name: RentsBySector
    removed: 2026-11-02
    use_instead: Rents
    note: split out of Rents; filter Rents on sector

Definitions

Key Type Required Priority Check Description
name string no None - Project name.
description string no Useful prose lint (warning) What the project is.
owner string no None - Who owns it.
license string no None - Licence of the data.
pricing list of PricingPlan no None see pricing Plans that apply to every dataset, unless a dataset declares its own.
facts list of Fact no by severity see facts Project-wide dated facts.
sources map of string to string no Important cross-project (see below) As-of source name to one line saying which vendor product or file it is. The same names are the keys of each dataset's dates_asof.
team string no None error unless it matches [A-Za-z0-9][A-Za-z0-9_.-]* Team that owns the project.
org string no None error unless it matches [A-Za-z0-9][A-Za-z0-9_.-]* Organisation that owns the project.
aliases list of string or {name, renamed} no Important error unless each name matches [A-Za-z0-9_.-]+; error if a name is also in retired_names Former project names that still resolve.
aliases[].name string yes (object form) Important as above The former name.
aliases[].renamed date YYYY-MM-DD no Important - When it was renamed.
retired_names list of object no Important see below Former names that no longer resolve.
retired_names[].name string yes Important error unless it matches the team pattern The retired name.
retired_names[].successor string no Important error unless it matches the team pattern The name that replaces it.
retired_names[].reason string yes Important - Why it was retired.
removed_datasets list of object no Important error if a name is still a dataset or a dataset's alias (a tool may report it as a warning while publishing) Datasets that used to exist (retired or split). Once a dataset is gone, this is the only place that says what replaces it.
removed_datasets[].name string yes Important as above The removed dataset.
removed_datasets[].removed date yes Important - When it was removed.
removed_datasets[].use_instead string no Important - The dataset (or <project>/<Dataset>) that replaces it.
removed_datasets[].note string no Important prose lint Free note, for example how to split the replacement.

Cross-project checks (when several projects are checked together): a source name used by one project that is an underscore-prefix of another project's source name (vendor and vendor_x) is an error, and the same source name described differently in two projects is a warning. Give one vendor file one shared name.

Four kinds of rename, at three levels. Project aliases: the project was renamed and the old name still works. Project retired_names: an old name that no longer works, with what replaced it. Dataset aliases: a dataset was renamed and the old name still works. Column aliases: a column was renamed and the old name still works. A dataset that is gone for good goes in removed_datasets. What to declare for each kind of change: Schema changes.