stack: Control-Plane-Sprache universell - Achse ist das Publikum, kein Instanz-Schalter (#103)
CI / verify (push) Failing after 45s
Release / release (push) Successful in 38s

Files changed:
- AGENTS.md
- CHANGES.md
- VERSION
- docs/language-boundaries.md
- kb/CONVENTIONS.md
- kb/CONVENTIONS.md.template
- tools/chemenu/commands/dist_cmd.py
This commit is contained in:
2026-09-15 16:58:55 +02:00
parent 05a75065ba
commit f350999053
7 changed files with 206 additions and 14 deletions
+22 -8
View File
@@ -107,15 +107,24 @@ documents. What it may not carry is the same content twice - a README that resta
contract is a second copy that drifts. `docs verify` enforces the specific case that already
happened once: no README may hold a copy of the `wikitool` command table.
**Two languages, and which is which.** Every file in the table above belongs to the stack and
ships to instances that share none of this instance's language choices, so:
**Two languages, and which is which.** Which one a line is written in follows from the *For*
column above - who reads it - and from nothing else: not from who owns the file, and not from
whether it ever leaves this checkout.
1. **The control plane is written in English** - this file, `CLAUDE.md`, every `CONTRACT.md`,
everything under `instructions/`, and the type-specs for non-page artifacts. Quoted
vocabulary is not prose and stays as it is: a section name, a relationship label or a
translated term cited as evidence. What the *instance* owns goes the other way - `kb/` pages
and the page type-specs that shape them follow `kb/CONVENTIONS.md`, which is also where the
instance's own terminology material is reached from.
translated term cited as evidence. What addresses the *page* goes the other way - `kb/` pages,
and inside a page type-spec the parts that become page text - and follows
`kb/CONVENTIONS.md`, which is also where the instance's own terminology material is reached
from.
This holds for a control-plane file an instance writes **only for itself** and never ships:
an instruction of its own, a page type it added (`types/` takes one without a code change),
a further stage contract. Such a file is instance-owned end to end, which settles who may
change it, not who reads it - and the reader is still an agent. There is deliberately no
second language value beside `kb/CONVENTIONS.md`'s `language:`, and no instance setting that
moves this rule; [docs/language-boundaries.md](docs/language-boundaries.md) has the reasoning.
2. **An agent speaks the instance's KB language**, whatever this file is written in. The value
lives in `kb/CONVENTIONS.md`'s `language:` and nowhere else; an instruction that models a
sentence for the user writes it in English like the rest of the control plane, and the agent
@@ -132,14 +141,19 @@ background consulted in passing, not a rule to follow; anything that would bind
type, index, lint or provenance; `dist export` ships it verbatim and no other `tools/wikitool`
command touches it.
Four pages exist today, each read by link rather than automatically:
Five pages are reached from this file, each by link rather than automatically:
[docs/pipeline-rationale.md](docs/pipeline-rationale.md) (why the pipeline has four stages),
[docs/ownership-and-templates.md](docs/ownership-and-templates.md) (why a `.template` split
exists, and why silent overwrite is the failure it guards against),
[docs/why-gates-are-code.md](docs/why-gates-are-code.md) (why the four gates in
[docs/language-boundaries.md](docs/language-boundaries.md) (why the control plane is English
everywhere and the KB language is a value, and why the axis is the reader rather than the
owner), [docs/why-gates-are-code.md](docs/why-gates-are-code.md) (why the four gates in
[Gates](#gates) are code rather than instruction), and
[docs/version-model.md](docs/version-model.md) (why a version number answers a compatibility
question and a migration question separately).
question and a migration question separately). A sixth,
`docs/model-and-effort-selection.md`, is deliberately not linked here but from `CLAUDE.md`: it
decides something only that harness has to decide, and a link here would load it into the other
three.
## Personalization
+34 -2
View File
@@ -59,7 +59,7 @@ concern - readable here, never shipped as something to parse.
---
## 6.0.0-beta.4 - 2026-09-15 - types/type-spec.md: Ownership und Sprache getrennt benannt (Nachzug zu #99)
## 6.0.0-beta.5 - 2026-09-15 - Control-Plane-Sprache universell: Achse ist das Publikum, kein Instanz-Schalter
**Author:** Torben Nehmer
@@ -77,6 +77,7 @@ concern - readable here, never shipped as something to parse.
**Medium impact**
- docs verify: ein nur als .template ausgeliefertes Linkziel gilt als aufgeloest
- Control-Plane-Sprache universell: Achse ist das Publikum, kein Instanz-Schalter
**Low impact**
- gates.md/session-setup.md: die Budget-Ausnahme von version regrade haengt an der Aufrufform
@@ -230,7 +231,38 @@ nicht haelt. Der Satz bindet ausdruecklich auch einen Type-Spec, den eine Instan
schreibt: der ist zwar durchgaengig instanzeigen, aber seine Anleitungshaelfte hat trotzdem einen
Agenten als Leser.
---
### Control-Plane-Sprache universell: Achse ist das Publikum, kein Instanz-Schalter
Die Sprachregel in `AGENTS.md` ruhte auf einer Begruendung, die schmaler war als sie selbst:
"Every file in the table above belongs to the stack and **ships to instances** that share none of
this instance's language choices, so:". Das traegt nur fuer ausgeliefertes Material und laesst
offen, was fuer ein Control-Plane-Dokument gilt, das eine Instanz nur fuer sich selbst schreibt -
eine eigene Instruction, ein selbst angelegter Seitentyp (`types/` nimmt einen ohne Code-Aenderung
auf), ein weiterer Stage-Contract. Genau dort fallen Ownership und Publikum auseinander: die Datei
ist durchgaengig instanzeigen, ihre Anleitungshaelfte hat trotzdem einen Agenten als Leser.
Der Vorsatz nennt jetzt die tatsaechliche Achse - die *For*-Spalte der Tabelle darueber, also wer
die Zeile liest, und weder wem die Datei gehoert noch ob sie den Checkout je verlaesst. Regel 1
sagt ausdruecklich, dass sie auch fuer ein nie ausgeliefertes Control-Plane-Dokument gilt und dass
es neben `kb/CONVENTIONS.md`s `language:` bewusst keinen zweiten Sprachwert gibt.
`kb/CONVENTIONS.md` und ihr `.template` sagen dasselbe von ihrer Seite aus: die
Control-Plane-Sprache ist keine Einstellung, die diese Datei zurueckhaelt - es ist gar keine.
Die Begruendung dazu steht als neue `docs/`-Seite
([docs/language-boundaries.md](docs/language-boundaries.md)), weil sie sonst in einem Jahr neu
verhandelt wird: warum Englisch (der Stack redet fast nur ueber Identifier, und die sind
englisch), warum kein Parameter (die Kosten traegt jede Datei, den Nutzen haette ein Dokument, das
ohnehin nur ein Agent liest), und was die Entscheidung wieder aufmachen wuerde. Die Seite haelt
zugleich fest, welches Argument falsch war: "Sprache folgt der Ownership" hat funktioniert,
solange nur ausgeliefertes Material betrachtet wurde, und faellt am instanz-eigenen Typ.
Nebenbei zwei Befunde derselben Ecke behoben. Der Docstring von
`dist_cmd.instance_owned_type_stems()` behauptete weiter, "its prose, its template and its
**language** are the instance's business" - Stand vor dem TOC-/Sprach-Bump oben. Und die
Aufzaehlung der `docs/`-Seiten in `AGENTS.md` sagte "Four pages exist today", waehrend das
Verzeichnis fuenf trug: `docs/model-and-effort-selection.md` fehlte, und zwar absichtlich, weil
ein Link dorthin die Claude-Code-eigene Entscheidung in die anderen drei Harnesses laden wuerde.
Der Satz zaehlt jetzt, was von hier aus verlinkt ist, und benennt die sechste Seite samt Grund.
## 5.1.0 - 2026-09-12 - changelog: Kandidaten-Eintrag nach Impact gruppiert, version regrade zur Nachkorrektur, version release verlangt eine Zusammenfassung
+1 -1
View File
@@ -1 +1 @@
6.0.0-beta.4
6.0.0-beta.5
+126
View File
@@ -0,0 +1,126 @@
# Language Boundaries
Two languages run through this repo at once. `kb/` is written in whatever language the instance
chose - German here, and the value lives in `kb/CONVENTIONS.md`'s `language:`. Everything that
tells an agent what to do - [AGENTS.md](../AGENTS.md), every `CONTRACT.md`, everything under
`instructions/` - is written in English, in every instance, whatever the first value says.
The rule itself is in [AGENTS.md § File naming](../AGENTS.md#file-naming). This page holds the
part that is not a rule: why the line runs where it does, why the English half is not a setting,
and which argument for it turned out to be wrong.
<!-- wikitool:toc -->
## Contents
- [The axis is the reader, not the owner](#the-axis-is-the-reader-not-the-owner)
- [Why the control plane's language is English](#why-the-control-planes-language-is-english)
- [Why it is not a parameter](#why-it-is-not-a-parameter)
- [What the KB language still decides](#what-the-kb-language-still-decides)
- [Where the line runs inside one file](#where-the-line-runs-inside-one-file)
- [What would put this back on the table](#what-would-put-this-back-on-the-table)
<!-- /wikitool:toc -->
## The axis is the reader, not the owner
For a long time the two halves could be told apart by asking who owned the file, and the answer
came out right every time: the stack owns `AGENTS.md` and the contracts, which are English; the
instance owns its pages and the templates that shape them, which are in the KB language. The
ownership boundary is a real and load-bearing thing - [ownership-and-templates.md](ownership-and-templates.md)
is about what it buys - so it was easy to read the language split as one of its consequences.
It is not. The case that separates them is a page type an instance adds for itself. `types/`
takes a new type without a code change, so an instance can write one; that file is instance-owned
from the first line to the last, ships nowhere, and is nobody's to overwrite. Its authoring
guidance is still instruction addressed to an agent, and reads exactly like the guidance in the
four types the stack ships. Ownership says "yours"; the audience has not moved at all.
So the question a line answers is not *whose file is this* but *who reads this line*, which is
the same cut [kb/CONTRACT.md](../kb/CONTRACT.md#language-and-identifiers) already makes inside a
single page between prose and identifiers - applied one level up, to the halves of a document.
Ownership decides who may change a sentence. The reader decides what language it is in. The two
questions were answered together for as long as they happened to agree.
## Why the control plane's language is English
Not because English is better for the purpose, and not to be neutral: this instance's operator
reads German, and the pages are German for that reason.
- **The control plane is almost entirely about identifiers, and the identifiers are English.**
`base_dir`, `provenance: sourced`, `--confirm`, exit 42, `root: kb`. A sentence in another
language explaining when to set `page_ref_fields` is already half English by the time it
reaches the verb, and the prose/identifier boundary inside it becomes something a reader has to
work out line by line.
- **It quotes a body of material that is English and stays English.** The harness documentation
it has to agree with, the vendored skill-authoring sources under `commonplace/`, the tool's own
`--help`. A contract that translates their vocabulary makes its own claims harder to check
against them, not easier.
- **One language keeps instances comparable.** Two instances running the same stack version hold
the same control plane byte for byte, so a question about one is answerable from the other -
and anything an instance changes locally shows up as a difference in content rather than in
language.
## Why it is not a parameter
The natural next move, once `kb/CONVENTIONS.md` holds `language:`, is a second value beside it -
`control_plane_language:` - defaulting to English and settable by an instance that would rather
read its contracts in its own language. That option is deliberately not taken.
- **The knob's cost is paid by every file; its benefit lands on the few a human reads.** Every
rule about writing an instruction would have to name which of the two languages it means, every
example would need a note saying which one it is in, and every review of an instruction would
start by establishing which language it should have been in. The stack has one mechanism for
that class of problem - one rule, one place (AGENTS.md invariant 8) - and a second language
value forks it everywhere at once.
- **The document the knob is for is read by an agent.** An instruction, a contract, a type-spec's
guidance half: the reader is a model, and a model reads the English fine. What the *operator*
reads is unaffected by any of this - see the section below.
- **Today's local document is tomorrow's upstream candidate.** An instruction an instance wrote
for itself is the most likely thing it ever contributes back. Written in the KB language it
would have to be translated first, and the translation would have to re-derive the
prose/identifier boundary that the original author had in their head and did not write down.
- **Nothing would check it.** There is no mechanical test for what language a paragraph is in -
a stop-word scan flags the quoted vocabulary the rule deliberately keeps and misses a cleanly
translated paragraph. A setting nothing enforces produces drift that is visible only to whoever
next opens the file.
## What the KB language still decides
Making the control plane English does not make the instance's language an implementation detail.
`kb/CONVENTIONS.md`'s `language:` decides two things, and both are the ones an operator actually
experiences:
- **Page text.** Every page under `kb/`, and inside the page type-specs exactly the parts that
become page text - each one's `## Template` block and its `layout:` titles.
- **What an agent says.** An agent speaks the KB language, whatever the file it just read was
written in. An instruction that models a sentence for the operator writes that model in
English, like the rest of the control plane, and the agent delivers it in the instance's
language.
So an operator who reads no English gets German pages and German answers from an agent reading
English instructions. The English is what the machinery is written in, not what it says back.
## Where the line runs inside one file
One file shape carries both halves, and it is worth knowing which is which before editing it: a
page type-spec (`types/entity.md` and its three siblings). Its authoring guidance addresses an
agent; its `## Template` block and `layout:` titles become the literal headings of pages; its
field names and enum values are identifiers and are translated in neither direction.
[types/type-spec.md § Who owns a type-spec](../types/type-spec.md#who-owns-a-type-spec) has that
split as a table, and [ownership-and-templates.md](ownership-and-templates.md) § "Where the file
boundary strains" has what it costs to keep two audiences in one file.
## What would put this back on the table
A `docs/` page goes stale when the reasoning stops holding rather than when the code changes, so
it is worth naming what that would look like here. Two things would:
- **A human starts reading the control plane directly and routinely** - not an operator checking
a rule now and then, which is the case today, but a workflow where people rather than agents
are the primary readers of `instructions/`. The second argument above is the one that fails
first, and it is the load-bearing one.
- **The identifiers stop being English.** If the tool's own vocabulary were ever localized, the
first argument would invert: the prose would then be the only English left in a file that is
otherwise not, which is the situation this page argues against.
Neither is close. Both are cheaper to notice here than to rediscover in an argument about a
single file.
+6
View File
@@ -53,6 +53,12 @@ Two things follow from that value rather than being decided here, both stated on
[AGENTS.md § File naming](../AGENTS.md#file-naming): the control plane stays English whatever an
instance writes its pages in, and an agent *speaks* the language named above.
The first of those is not a setting this file withholds - it is not a setting at all. `language:`
above is the only language value in the tree, and what it binds is page text; a control-plane
document is English even when this instance wrote it for itself and never ships it. Why that is
an architecture decision rather than an unset parameter:
[docs/language-boundaries.md](../docs/language-boundaries.md).
Which line is prose and which is an identifier - and therefore what is translated at all - is
the contract's rule, not this file's: see
[kb/CONTRACT.md § Language and identifiers](CONTRACT.md#language-and-identifiers).
+6
View File
@@ -42,6 +42,12 @@ Two things follow from that value rather than being decided here, both stated on
[AGENTS.md § File naming](../AGENTS.md#file-naming): the control plane stays English whatever an
instance writes its pages in, and an agent *speaks* the language named above.
The first of those is not a setting this file withholds - it is not a setting at all. `language:`
above is the only language value in the tree, and what it binds is page text; a control-plane
document is English even when this instance wrote it for itself and never ships it. Why that is
an architecture decision rather than an unset parameter:
[docs/language-boundaries.md](../docs/language-boundaries.md).
Which line is prose and which is an identifier - and therefore what is translated at all - is
the contract's rule, not this file's: see
[kb/CONTRACT.md § Language and identifiers](CONTRACT.md#language-and-identifiers).
+11 -3
View File
@@ -308,9 +308,17 @@ def instance_owned_type_stems() -> set[str]:
The line is `root:`, and it was already in the frontmatter before anyone
drew it: `root: kb` means the type describes a page the instance writes, so
its prose, its template and its language are the instance's business.
Anything else - `instruction` (`root: repo`), `lint-report` (no `base_dir`
at all), `type-spec` itself - describes a stack artifact and ships verbatim.
the file is the instance's to change. Anything else - `instruction`
(`root: repo`), `lint-report` (no `base_dir` at all), `type-spec` itself -
describes a stack artifact and ships verbatim.
Ownership, not language. What such a file is *written in* is decided by who
reads each half, not by who owns the file: its `## Template` block and its
`layout:` titles become page text and follow `kb/CONVENTIONS.md`, while the
authoring guidance around them addresses an agent and stays English like the
rest of the control plane (AGENTS.md § File naming, types/type-spec.md
§ Who owns a type-spec). That the two halves share one file, and what it
costs, is docs/ownership-and-templates.md § Where the file boundary strains.
Read from `types/` rather than listed, so an instance adding its own page
type gets the same treatment without a code change.