docs(adr-0036): re-anchor the enforcement-rule citation to the framework's ADR-0124 D1 (#5701) [GOVERNED SURFACE — human merge]

[os-zhuang]

Refs #5701

⛔ GOVERNED SURFACE — this PR touches docs/adr/**. It stops at draft for human merge.
Do not flip it ready, do not enqueue it, do not enable auto-merge. It waits on a human's
schedule by design.

Re-anchors the enforcement-rule citation in docs/adr/0036-field-conditional-rules.md from
the framework's ADR-0057 D10 to the framework's ADR-0124 D1.

Why this is a separate PR

This paragraph is the wording every other citation of the rule in this repository derives
from — so leaving it in place re-seeds the misattribution into anything written from it
next. That makes it required work, not optional.

But docs/adr/** is governed, and a PR touching it merges on a human's schedule. Bundling
it with the eight ordinary code sites would make those eight wait on a human for no reason,
so the sweep splits: the code sites ship in their own PR (see below), and the ADR waits
here.

The change

One paragraph, docs/adr/0036-field-conditional-rules.md:89-94.

Before

Enforcement belongs on the server — server enforces, client is courtesy, the rule the
framework cites throughout its rule-validator, its lint diagnostics and its QA runner
as ADR-0057 D10 (that is the framework's ADR numbering; this repo's own
ADR-0057, 0057-console-ai-chat-one-conversation-docked.md, is an unrelated document) —
and a declared lock that failed open would leave enforcement in the courtesy layer
instead.

After

Enforcement belongs on the server — server enforces, client is courtesy, the rule the
framework decides in ADR-0124 D1 (framework ADR numbering) and practises
throughout its rule-validator, its lint diagnostics and its QA runner — and a declared
lock that failed open would leave enforcement in the courtesy layer instead.

Three deliberate choices in that rewrite:

1. "cites … as" → "decides in … and practises throughout".ADR-0124 D1 is the record
   that decides the rule; ADR-0057 D10 was only ever cited for it. Keeping "cites as"
   with the new number would swap one inaccuracy for another.
2. The rule-validator / lint-diagnostics / QA-runner triple is unchanged — and is not a
   guess: framework ADR-0124's own Consumers line names validation/rule-validator.ts,
   @objectstack/lint, and docs/qa/platform-checklist (RUNNER rule 4).
3. The 0057-collision parenthetical retires with the number it disambiguated. It
   warned that this repo's own ADR-0057 is an unrelated document; this repo has no
   ADR-0124 at all (its own series stops at 0059), and framework ADR-0124 records that
   a fresh, unambiguous number was chosen precisely so its citations would not need such a
   warning. (framework ADR numbering) keeps the whose-numbering half.

Why the old anchor is wrong

Framework ADR-0057D10 decides "Setup-nav surfacing follows the capability
(ADR-0029 K2); the object stays open" — nav-entry tiering, not enforcement location. The
rule this paragraph asserts is decided by framework ADR-0124 D1, "The server is the
enforcement point; client-side gating is a usability courtesy" (Accepted 2026-08-18).

The new anchor is derived from an authority, not chosen: framework ADR-0057 carries a
note aimed at exactly this citation — "If a citation of ADR-0057 D10 brought you here
looking for that rule, ADR-0124 is where it is decided (#9628)."

Verification — all at b6fc3b5f5

Check	Result (the gate's own verdict line)
check-doc-links.mjs	Links are valid across 13 scan roots.
check-control-bytes.mjs	✅ check-control-bytes: OK (scanned 4855 tracked text file(s); skipped 85 binary)
check-changeset-presence.mjs	✅ No source of a released package changed in this range, so no changeset is owed.
pnpm exec vitest run scripts/__tests__/doc-version-claims.test.ts	Test Files 1 passed (1) · Tests 18 passed (18)

The changeset gate was run, not assumed — it reports that a docs-only diff owes nothing,
so no changeset is added. doc-version-claims guards against a bare version literal reaching
prose; none was written.

Prose-only change, one file, +4 / −5. Nothing to ablate — there is no behavioural leg
here, so no mutation could make a test fail differently; a staged reverse-verification would
be theatre.

Companion PR

The eight code sites that derive their wording from this paragraph move in their own,
ordinary PR — it carries the full before/after enumeration showing that once both land, the
only surviving live-source ADR-0057 D10 citation in this repo is
packages/data-objectstack/src/appAccessProbe.test.ts, which cites it for the
capability/service-gating case the decision genuinely does decide.

---

Generated by Claude Code