railiance-master/docs/adr/ADR-0010-taxonomy-is-an-enumerated-layer.md
codex d7a95a2d50 Apply GH-DEC-2026-017 to the Taxonomy layer declaration and close RMASTER-WP-0027.
INTENT.md governs; layer.yaml is marked derived from it. standard_version is
removed from both. The value Taxonomy is unchanged: the ruling closes the
vocabulary at four tokens and Taxonomy is one of them. ADR-0010 amended by
revision (accepted-2). No checker in this repository reads the declaration.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 63291@bnt-lap001
Assistant-Session: 8bd77868-ca68-4f49-bb1e-d539ecc0d703
2026-09-21 07:33:58 +02:00

196 lines
10 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

---
id: RMASTER-ADR-0010
title: "Taxonomy Is An Enumerated Layer, Not A Value Outside The Vocabulary"
status: accepted
revision: "accepted-2"
owner: railiance-master
last_reviewed: "2026-09-21"
review_interval: 6m
---
# ADR-0010: Taxonomy Is An Enumerated Layer, Not A Value Outside The Vocabulary
Date: 2026-09-21
Status: Accepted
## Context
`flex-auth` published a boundaries review (`FLEX-WP-0030`,
`flex-auth/docs/conformance/boundaries-review.md`) on 2026-09-20, corrected
2026-09-21, surveying how repositories declare their layer under §11 of the
NetKingdom security layer model. It surveyed the twelve counterparts
`flex-auth` holds boundaries with; this repository is not one of them and was
not surveyed. The custodian then swept the whole estate and recorded the result
in `the-custodian/docs/assessments/2026-09-21-layer-declaration-boundaries.md`.
That sweep is where this reached us. It was never a finding against this
repository, and nobody had raised it here before.
The sweep records that `flex-auth`'s validator admits only
`{Staff, Engine, Tooling}`, and that two repositories declare values outside
that set: `informed-decision` declares `surface`, and this repository declares
`Taxonomy`. It classes both as a third non-conformance class — "the only one
where the declared layer is not a layer the model defines" — and refers the
question of whether §3's vocabulary is closed to `gate-house`.
This ADR is this repository's input to that ruling. It does not rule; §11 and
§3 are net-kingdom canon authored by `gate-house`, and ADR-0009 already binds
us to that division.
## Decision
1. **`Taxonomy` is a layer the model defines.** §3 of the security layer model
enumerates four layers, and `Taxonomy` is the first row of its own table.
§3.1 is the section that defines it. This is true of every version of the
standard from v0.1 through v0.8, the proposed version under assent:
| Layer | Character | Produces | Deterministic |
| --- | --- | --- | --- |
| **Taxonomy** | cross-cutting language | terms, semantic contracts, standards | n/a — describes |
| **Tooling** | infrastructure and state | data structures, persistence | yes |
| **Engines** | interfaces for a modeled concept | APIs, contracts | yes |
| **Staff** | management, operations, change, controlling | specifications, decisions, workplans, tasks | **no** |
The §3 vocabulary is therefore not `{Staff, Engine, Tooling}` and never was.
A validator admitting those three omits §3.1.
2. **We do not change our declared value.** `layer: Taxonomy` stands in both
`INTENT.md` frontmatter and `layer.yaml`. The artifact that disagrees with
§3 is the validator's admitted set, not our declaration.
3. **We take no position on `surface`.** `informed-decision`'s value is a
genuinely different question: `surface` does not appear in §3 under any
version. Grouping it with `Taxonomy` as one finding is the part of the
sweep's third class we do not accept. That repository speaks for itself.
4. **We take no position on the case-sensitivity question (B1/question 2)**
beyond the observation in the evidence below, and none on precedence
(question 1), which does not bite here: our two files agree.
5. **If `gate-house` rules otherwise** — that §3's vocabulary is closed at
three for §11 declaration purposes and `Taxonomy` is not a declarable value
— we will change the declaration in both files in one commit and amend this
ADR by revision, not by silent edit. We will not pre-empt that ruling by
changing the value now, because the declaration is the evidence of what this
repository holds itself to be.
## Evidence
The term is load-bearing in canon and in live estate use, not a local coinage:
- **§3 and §3.1** of `security-layer-model` v0.1 through v0.8 define the
Taxonomy layer: "Cross-cutting language. Taxonomy repositories define terms
and semantic contracts so the other layers interoperate without integration
by interpretation. They own no runtime position and no state any layer
depends on."
- **§4**, the security-estate catalog, assigns the layer `Taxonomy` to two
rows in v0.8: `info-tech-canon` and `net-kingdom`. The model uses the value
as a layer assignment in its own catalog.
- **§7** maps the layers onto the Active Secrets Management Canon planes and
carries `Taxonomy` as a fourth line alongside Staff, Engines and Tooling.
- **§17 is titled "Taxonomy artifacts"** and assigns ownership of the claim,
decision-record, gap-record and stance-map schemas to "Taxonomy" as a layer
with owners.
- **`gate-house`'s own `INTENT.md`** (lines 8788) records `info-tech-canon`
and `net-kingdom canon` in the `Taxonomy` layer. The doctrine owner uses the
value.
- **`info-tech-canon`'s `INTENT.md`** declares itself "part of the **Taxonomy**
layer" in prose. It carries no `layer:` key and no `layer.yaml`, so it is a
§11 B2 case rather than a vocabulary case — but its intent is the same value.
- **Our own casing matches §3's table.** `Taxonomy` is capitalised in §3 as
`Tooling`, `Engines` and `Staff` are. The nine repositories whose `layer.yaml`
lower-cases `engine` or `staff` are the ones that diverge from the standard's
own spelling; we do not. We also note the validator admits `Engine` where §3
writes `Engines`, which suggests the admitted set was built from §4's
role-typed catalog rows rather than from §3.
## What `Taxonomy` denotes here, and why not Staff, Engine or Tooling
This repository produces terms, family contracts and standards for Railiance
workload operations — the `railiance-*` / `rail-*` / `rapp-*` / `reef-*` axes,
the workload coverage rule, and the composition semantics. That is §3.1's
character exactly, and `layer.yaml` states §3.1's own test in §3.1's own words:
it owns no runtime position and no state another layer depends on.
- **Not Engine.** It exposes no deterministic API over a modeled concept, is
not PEP-shaped, and ADR-0009 §5 and INTENT principle 13 forbid it becoming a
second decision point. `role: null` follows.
- **Not Tooling.** It persists nothing and keeps no state.
- **Not Staff.** Staff is management, operations, change and controlling. This
repository defines vocabulary and boundaries; the rails, rapps and reefs that
operate workloads are elsewhere and carry the §6.4 obligations themselves.
The usage is consistent with `gate-house`'s and `info-tech-canon`'s: the same
layer, a different scope. `layer.yaml`'s `catalog_entry.note` has stated that
distinction since 2026-08-29 — "This is Railiance operations Taxonomy, not
NetKingdom security Taxonomy. info-tech-canon remains ecosystem-wide semantics;
net-kingdom remains NetKingdom standards of record" — and states in the same
breath that we are **not** a §4 catalog row. Declaring the Taxonomy layer and
claiming a §4 row are two different claims, and we make only the first.
The value was chosen deliberately, under `RMASTER-WP-0026-T01`, which
instructed "keep the file and the frontmatter equal". Internal agreement here
is a decision that was taken, not an accident of a generator.
## Consequences
- `gate-house` has, in this ADR, a citable statement that the value under
question is enumerated in the section said to exclude it. The cheapest
resolution of question 3 is a correction to the validator, not to two
repositories' declarations.
- The custodian's third non-conformance class splits. `Taxonomy` is a validator
defect; `surface` remains open on its own terms.
- Our `standard_version: "0.7"` in both files is a separate question — B4 asks
whether a declaration may carry a standard version at all. We hold it
unchanged pending that ruling for the same reason as the layer value, and
will re-point at the accepted version once v0.8's assent round closes.
- ADR-0009 is not amended. This ADR adds the vocabulary position that record
did not have to state, because nothing had questioned it.
## Notes
Tracked in `RMASTER-WP-0027`. Sent to `gate-house` as input to the §3
vocabulary ruling on 2026-09-21.
## Revision accepted-2 (2026-09-21) — the ruling
`gate-house` ruled in `GH-DEC-2026-017` (gate-house `decisions/decisions.md`;
amendments A9, A11 and A12 in
`gate-house/docs/amendments/v0.8-section-11-declaration-amendments.md`). The
text above is left as accepted-1 recorded it; this section states what the
ruling confirmed and what it changed.
**Confirmed, unchanged:**
- Decision 1 and 2. The §3 vocabulary is closed at **four** tokens —
`Taxonomy`, `Tooling`, `Engine`, `Staff` (§3 of the ruling, A9).
`layer: Taxonomy` is conforming and the validator carried the defect.
`flex-auth` has since fixed it (flex-auth `80ffe72`). The value is not
changed; the contingency in Decision 5 did not trigger.
- Decision 3. `surface` was ruled apart from `Taxonomy`, as asked.
- Scope. This repository is outside §4. The ruling (§4, A11) holds that §11
binds §4 repositories only, so our declaration is *declared voluntarily,
outside catalog scope*. It was never a §11 non-conformance, and is not one
now.
**Changed by the ruling:**
- **Precedence (Decision 4, question 1).** `INTENT.md` frontmatter governs.
`layer.yaml` is a derived artifact and now carries `derived: true` and
`derived_from: INTENT.md`. The two still agree.
- **Case (Decision 4, question 2; the casing observation under Evidence).**
Comparison is ASCII case-insensitive and a lowercase declaration is
conforming, not tolerated. The Evidence observation that the nine
lower-casing sidecars "diverge from the standard's own spelling" is
superseded: under A9 they conform. Its underlying point — §3 writes
`Engines`, §4 writes `Engine` — became A9, which states the token once and
makes the §4 column spelling canonical. Our `Taxonomy` is canonical under
either reading.
- **Standard version (Consequences, third bullet).** A layer declaration MUST
NOT carry a standard version (§5 of the ruling, A12). `standard_version: "0.7"`
is removed from `INTENT.md` frontmatter and `layer.yaml`, not re-pointed at
v0.8 as that bullet anticipated. Version-scoped state belongs in a derived
conformance record; this repository keeps none.
This repository has no layer conformance checker of its own, so no checker or
test had to change with the field. `RMASTER-WP-0027-T03` closes on this
revision.