Browse the docs

DocsReference

Ekwo maintains a pack; only an accountant reviews one

Status: accepted

Context

A golden test proves a pack is internally coherent. Neither writing a pack nor testing it is an accountant reading it against the law, so "certified by Ekwo" is a claim the project cannot make.

Decision

The scale is community, maintained, reviewed.

  • community — contributed, not read by an accountant.
  • maintained — kept current by the maintainers, not yet reviewed by an accountant.
  • reviewed — a named professional read it; the manifest carries who and when. Nothing else counts.

There is no value meaning "certified by Ekwo". A deprecated enum value that once meant so remains in the enum (a published column never loses a value), nothing writes it, the pack schema refuses it, and rows holding it were migrated.

One sentence describes a pack, written once in the CLI and printed by ekwo init, ekwo status and every generated seed header: "maintained by Ekwo — not yet reviewed by an accountant", "reviewed by X on Y", "community pack — not reviewed".

A pack that is not reviewed says where a reviewer should start. Its README lists the points where the text allows more than one reading. A list of known soft spots is worth more than a status that overstates the work.

legal_reference is required on every tax and every box, and a "TODO" is not a source. .github/CODEOWNERS names who is asked about each pack — not who has signed it.

A whole chart, at the depth the law prescribes. Where a country publishes a chart, the pack transcribes it whole rather than ship a subset, which would be this repository's opinion of which accounts matter; where the State publishes the mapping from accounts to statement lines, the pack uses it account by account rather than compressing it into ranges.

A country written from published sources alone tests the format. A pack written by somebody reading a foreign statute with no permission to change the core is the honest test that a country is data.

Consequences

  • No status in the product claims more review than happened.
  • Moving a pack to reviewed is a manifest change signed by a named person.

See also

Rendered from docs/decisions/0026-certification-names-a-reviewer.md, in the repository, when the site was built. Read or change it on GitHub