diff --git a/.changeset/9941-tabs-default-items-content.md b/.changeset/9941-tabs-default-items-content.md
new file mode 100644
index 0000000000..83a8346a0b
--- /dev/null
+++ b/.changeset/9941-tabs-default-items-content.md
@@ -0,0 +1,23 @@
+---
+'@object-ui/components': patch
+---
+
+The shipped `ui:tabs` registration's `defaultProps.items` now spell `content`, the
+child key the published `TabItemSchema` declares, instead of the undeclared `body`
+(objectui#9941).
+
+All three seeded items spelled `body` and omitted `content`, so the default an
+author drops on a canvas was refused by the validator this repository publishes
+for it — `invalid_type ["content"] expected nonoptional, received undefined`,
+measured against the built schema, with the lit control that the same item spelled
+`content` parses green with keys `value,label,content`.
+
+⚠️ No rendered output moves. `tabs.tsx` reads `item.content` first and falls back
+to `body` through an `any` cast, so the same nodes move from the fallback arm onto
+the primary one; the repair is at the parse and the rendered markup is unchanged
+(asserted, not assumed). What changes for an author is that copying the shipped
+default into authored metadata now validates.
+
+The fallback arm itself is untouched and `TabItemSchema` is untouched — widening
+the published accept set to admit `body` would pre-empt the `body`-dialect
+question open on objectui#9871.
diff --git a/packages/components/src/renderers/layout/__tests__/tabs-default-items-parse-9941.test.tsx b/packages/components/src/renderers/layout/__tests__/tabs-default-items-parse-9941.test.tsx
new file mode 100644
index 0000000000..36dbff3140
--- /dev/null
+++ b/packages/components/src/renderers/layout/__tests__/tabs-default-items-parse-9941.test.tsx
@@ -0,0 +1,230 @@
+/**
+ * ObjectUI
+ * Copyright (c) 2024-present ObjectStack Inc.
+ *
+ * This source code is licensed under the MIT license found in the
+ * LICENSE file in the root directory of this source tree.
+ */
+
+/**
+ * The shipped `ui:tabs` registration's `defaultProps.items` satisfy the
+ * published `TabItemSchema` (objectui#9941).
+ *
+ * ## The defect this pin was built around
+ *
+ * `TabItemSchema` (`packages/types/src/zod/layout.zod.ts`, cited by SYMBOL —
+ * objectui#8875) declares `content` with **no** `.optional()` — required — and
+ * declares **no `body` member at all**. The registration seeded three items
+ * that spelled `body` and omitted `content`, so the default an author drops on
+ * a canvas was refused by the validator this repository publishes for it:
+ *
+ *     invalid_type ["content"] expected nonoptional, received undefined
+ *
+ * ⚠️ The defect was never visible in the rendered UI. `tabs.tsx` reads
+ * `renderChildren(item.content || (item as any).body)` — `content` first, `body`
+ * through an `any` cast — so the seeded nodes drew from the fallback arm. It was
+ * the DECLARED contract that refused them, and only a parse can see that.
+ *
+ * ## Why this pin parses instead of reading the spelling
+ *
+ * A test asserting the three items spell `content` is satisfied forever by a
+ * copy-paste of the literal; it re-states the fix instead of checking it. This
+ * one feeds whatever the registry actually carries through the published schema,
+ * so it fails the day anyone re-spells a member, adds a fourth item that omits
+ * `content`, or moves the seed — and it fails for the same reason the card was
+ * filed, not for a reason a reader has to reconstruct.
+ *
+ * Both faces are read off live artifacts: the seed off `ComponentRegistry` (the
+ * object a designer palette consumes), the contract off `@object-ui/types/zod`
+ * (the object a validator consumes). Nothing below is a copy of either.
+ *
+ * ## Which rows discriminate, and which are controls
+ *
+ * ⚠️ The `controlB` rows are green in BOTH worlds BY DESIGN — that is the whole
+ * claim of this card. The repair is at the parse, so a row that moved with it
+ * would be evidence the repair was visual after all. The rows that discriminate
+ * are the three subject rows; `controlA` and `controlB` exist to prove the
+ * instrument can fail and that the UI did not follow, respectively.
+ *
+ * ## The instrument's two failure modes, each with its own control
+ *
+ * 1. **A parse over nothing is green.** An empty (or missing) `items` array
+ *    makes `every item parses` vacuously true, so the seed's presence and
+ *    non-emptiness are asserted first, as their own rows.
+ * 2. **A parse that cannot refuse anything is green.** `controlA` feeds the
+ *    pre-fix item — the same objects with `content` re-spelled back to `body` —
+ *    to the same call and requires a refusal that names `content`. If that row
+ *    goes green the subject rows below measure nothing.
+ *
+ * ## ⛔ What this pin deliberately does NOT claim
+ *
+ * - ⛔ It does not assert that `body` is refused. `TabItemSchema` is a stripping
+ *   `z.object`, so an undeclared `body` is silently DROPPED, not rejected;
+ *   `undeclaredBodyIsDropped` records that as the measured behaviour rather
+ *   than letting a later reader assume a strict face.
+ * - ⛔ It does not touch the `|| (item as any).body` fallback in `tabs.tsx`.
+ *   Retiring that tolerance is the `body`-dialect family's question
+ *   (objectui#9871, objectui#9910 — both open, both `needs-user-decision`), and
+ *   a pin here must not answer it. ⚠️ `controlB.equality` below is the one row
+ *   that READS that arm: it exists to prove this repair was parse-level and not
+ *   visual. When the fallback is retired, that row retires WITH it — ⛔ it is
+ *   not a reason to weaken the subject, and the durable half of control B
+ *   (`controlB.stillRenders`) does not depend on the arm at all.
+ * - ⛔ It is not widened past `ui:tabs`. Whether a registration's `defaultProps`
+ *   may diverge from its published face across the repository is the open
+ *   question on objectui#4631 (`pm:on-hold`), and deciding it by gate rather
+ *   than by ruling is the trap the sibling pin in this directory
+ *   (`registration-defaults-match-renderer-8229.test.ts`) documents at length.
+ */
+
+import { describe, it, expect, afterEach } from 'vitest';
+import { render, cleanup } from '@testing-library/react';
+import { ComponentRegistry } from '@object-ui/core';
+// Registered at module scope, NOT in a hook: a cold transform billed to
+// `hookTimeout` is narrower than the timeout it replaces (objectui#3010).
+import '../../../renderers';
+import { TabItemSchema, TabsSchema } from '@object-ui/types/zod';
+
+afterEach(() => cleanup());
+
+type Item = Record<string, unknown>;
+
+const meta = () => ComponentRegistry.getMeta('tabs', 'ui');
+const seed = () => (meta()?.defaultProps ?? {}) as Record<string, unknown>;
+const seededItems = () => (seed().items ?? []) as Item[];
+
+/**
+ * The pre-fix spelling, rebuilt from whatever the registry carries today: the
+ * item's child list, wherever it is spelled, moved onto `body` with `content`
+ * gone. Reading BOTH spellings keeps this a faithful inverse under ablation —
+ * re-spell the seed back to `body` and this returns the seed unchanged, so the
+ * control rows stay controls and only the subject rows move.
+ */
+const preFix = (items: Item[]): Item[] =>
+  items.map(({ content, body, ...rest }) => ({ ...rest, body: content ?? body }));
+
+/** Render the registration seed as a `tabs` node and return what it paints. */
+function renderSeed(items: Item[]): { text: string; html: string } {
+  const C = ComponentRegistry.get('tabs', 'ui');
+  if (!C) throw new Error('no renderer registered for ui:tabs');
+  render(<C schema={{ type: 'tabs', ...seed(), items }} />);
+  // Radix mints a fresh id per mount, so two renders of identical content
+  // differ in `id`/`aria-controls`/`aria-labelledby` and nowhere else. Blank
+  // those three out; everything that carries meaning survives.
+  const html = (document.body.innerHTML ?? '').replace(
+    /(id|aria-controls|aria-labelledby)="[^"]*"/g,
+    '$1="#"',
+  );
+  return { text: document.body.textContent ?? '', html };
+}
+
+describe('`ui:tabs` defaultProps.items satisfy the published TabItemSchema (objectui#9941)', () => {
+  describe('the seed is really there — ⛔ a parse over nothing is vacuously green', () => {
+    it('the registration is registered and carries a defaultProps seed', () => {
+      expect(meta(), 'ui:tabs is not registered — every row below is vacuous').toBeDefined();
+      expect(meta()?.defaultProps).toBeDefined();
+    });
+
+    it('the seed carries a non-empty `items` array', () => {
+      expect(Array.isArray(seed().items)).toBe(true);
+      expect(seededItems().length).toBeGreaterThan(0);
+    });
+  });
+
+  describe('controlA — the instrument can FAIL', () => {
+    it('the pre-fix item is refused, and the refusal names `content`', () => {
+      // The exact reading objectui#9941 was filed on, re-taken through the same
+      // call the subject rows use. Structured fields only — ⛔ no message prose.
+      const refused = preFix(seededItems()).map((item) => TabItemSchema.safeParse(item));
+      expect(refused.length).toBeGreaterThan(0);
+      for (const r of refused) {
+        expect(r.success).toBe(false);
+        if (r.success) continue;
+        expect(r.error.issues.map((i) => ({ code: i.code, path: i.path.join('.') }))).toContainEqual({
+          code: 'invalid_type',
+          path: 'content',
+        });
+        const missing = r.error.issues.find((i) => i.path.join('.') === 'content');
+        expect((missing as { expected?: string } | undefined)?.expected).toBe('nonoptional');
+      }
+    });
+
+    it('undeclaredBodyIsDropped — `body` is stripped by the parse, ⛔ not rejected', () => {
+      // Why the refusal above is about the ABSENT key and never about the
+      // present one: the face is a stripping `z.object`, so the item's own
+      // child list did not survive the parse even before the missing-key error.
+      const parsed = TabItemSchema.parse({
+        value: 'probe',
+        label: 'probe',
+        content: [{ type: 'text', content: 'kept' }],
+        body: [{ type: 'text', content: 'dropped' }],
+      });
+      expect(Object.keys(parsed).sort()).toEqual(['content', 'label', 'value']);
+      expect(parsed).not.toHaveProperty('body');
+    });
+  });
+
+  describe('the subject', () => {
+    it('every seeded item parses green against `TabItemSchema`', () => {
+      const failures = seededItems()
+        .map((item, i) => ({ i, r: TabItemSchema.safeParse(item) }))
+        .filter(({ r }) => !r.success)
+        .map(({ i, r }) =>
+          `items[${i}]: ${(r as { error: { issues: { code: string; path: PropertyKey[] }[] } }).error.issues
+            .map((issue) => `${issue.code} [${issue.path.join('.')}]`)
+            .join(', ')}`,
+        );
+      expect(
+        failures,
+        'the shipped default is the metadata an author drops on a canvas — it has to ' +
+          'satisfy the schema this repository publishes for it. `TabItem`\'s child key is ' +
+          '`content` (⛔ not `body`, ⛔ not `children`); spell the seed that way rather than ' +
+          'widening `TabItemSchema`, which would grow an already-published accept set and ' +
+          'pre-empt objectui#9871.',
+      ).toEqual([]);
+    });
+
+    it('the whole registration seed parses as a `tabs` node', () => {
+      // One step above the item face: what the designer writes out is the node,
+      // not a loose item. Green here is the end-to-end statement of the repair.
+      const r = TabsSchema.safeParse({ type: 'tabs', ...seed() });
+      expect(r.success, JSON.stringify(r.success ? [] : r.error.issues, null, 1)).toBe(true);
+    });
+
+    it('each seeded item keeps a child list after the parse', () => {
+      // Green-because-empty guard on the row above: an item whose `content` was
+      // an empty array would parse and carry nothing.
+      for (const item of seededItems()) {
+        const parsed = TabItemSchema.parse(item) as { content: unknown };
+        expect(parsed.content).toBeDefined();
+        expect(Array.isArray(parsed.content) ? (parsed.content as unknown[]).length : 1).toBeGreaterThan(0);
+      }
+    });
+  });
+
+  describe('controlB — the UI did not move', () => {
+    it('stillRenders — the seed paints every label and the first tab body', () => {
+      // Durable half: reads only the `content` arm, so it survives whatever
+      // objectui#9871 rules about the fallback.
+      const { text } = renderSeed(seededItems());
+      for (const item of seededItems()) expect(text).toContain(String(item.label));
+      expect(text).toContain('Content for Tab 1');
+    });
+
+    it('equality — the pre-fix spelling paints exactly the same thing today', () => {
+      // ⚠️ THIS ROW READS THE FALLBACK ARM (`item.content || (item as any).body`).
+      // It is the proof that objectui#9941 was a parse-level repair and not a
+      // visual one: the same nodes moved from the fallback arm onto the primary
+      // one. It is ⛔ NOT a claim that the fallback must stay — when
+      // objectui#9871 retires it this row goes with it, and `stillRenders`
+      // above, plus every subject row, are untouched by that.
+      const after = renderSeed(seededItems());
+      cleanup();
+      const before = renderSeed(preFix(seededItems()));
+      expect(before.text).toBe(after.text);
+      expect(before.html).toBe(after.html);
+      // Lit control: ⛔ two empty renders are also equal.
+      expect(after.text).toContain('Content for Tab 1');
+    });
+  });
+});
diff --git a/packages/components/src/renderers/layout/tabs.tsx b/packages/components/src/renderers/layout/tabs.tsx
index 502350f2bc..cfa19819de 100644
--- a/packages/components/src/renderers/layout/tabs.tsx
+++ b/packages/components/src/renderers/layout/tabs.tsx
@@ -85,9 +85,9 @@ ComponentRegistry.register('tabs', TabsRenderer,
     defaultProps: {
       defaultValue: 'tab1',
       items: [
-        { label: 'Tab 1', value: 'tab1', body: [{ type: 'text', content: 'Content for Tab 1' }] },
-        { label: 'Tab 2', value: 'tab2', body: [{ type: 'text', content: 'Content for Tab 2' }] },
-        { label: 'Tab 3', value: 'tab3', body: [{ type: 'text', content: 'Content for Tab 3' }] }
+        { label: 'Tab 1', value: 'tab1', content: [{ type: 'text', content: 'Content for Tab 1' }] },
+        { label: 'Tab 2', value: 'tab2', content: [{ type: 'text', content: 'Content for Tab 2' }] },
+        { label: 'Tab 3', value: 'tab3', content: [{ type: 'text', content: 'Content for Tab 3' }] }
       ],
       className: 'w-full'
     }
diff --git a/scripts/__tests__/body-dialect-census.test.ts b/scripts/__tests__/body-dialect-census.test.ts
index 86d3c9dd7f..894fd5c7a7 100644
--- a/scripts/__tests__/body-dialect-census.test.ts
+++ b/scripts/__tests__/body-dialect-census.test.ts
@@ -374,8 +374,23 @@ describe('the `body` consumers the ruling does not enumerate', () => {
     // rather than reasoned, both legs, against the built parser: a `card` node
     // carrying `body` draws `unknown-prop: <card> has no prop "body" — the
     // child-list key is "children"`, and a `tabs` item carrying the same key
-    // draws ZERO diagnostics. ⇒ an empty table is objectui#9590's finish line,
-    // ⛔ not step 4's gate. ⛔ Do not delete the claim to make the block green.
+    // draws ZERO diagnostics.
+    //
+    // ⭐ RE-POINTED at objectui#9941, ⛔ not deleted — the assertion below
+    // carried its own instruction to do exactly that, in these terms. The table
+    // emptied when the three `tabs.tsx` `defaultProps` items were respelled to
+    // `content`, the key `TabItemSchema` declares required while declaring no
+    // `body`: ⛔ a CONFORMANCE fix, ⛔ not a dialect migration.
+    //
+    // ⭐ And the re-pointed claim is SCOPED: the table is empty **under
+    // objectui#9871's EMISSION criterion**. ⛔ It is NOT a claim that the
+    // dialect has left the tree, and ⛔ NOT objectui#9590's finish line.
+    // Measured across the same change: `failedC2` held at 313 and
+    // `unclassified` at 5, so 318 rows outside what that criterion scores as
+    // emission did not move — and objectui#9989 is a published doc still
+    // TEACHING the spelling. ⭐ Emitted and taught are different verbs, and this
+    // scan reads only the first. Whether the finish line is reached is
+    // objectui#9590's to judge on its own record.
     const producerScan = scan(REPO_ROOT);
 
     // Two lit controls, because an empty table is the assertion's own shape: a
@@ -391,20 +406,30 @@ describe('the `body` consumers the ruling does not enumerate', () => {
     const producers = producersOf(producerScan.hits, producerScan.readers);
     expect(
       producers.length,
-      'the producer table is EMPTY — under objectui#9871\'s criterion nothing in shipped source ' +
-        'emits the dialect any more. ⚠️ That is objectui#9590\'s finish line, ⛔ NOT step 4\'s gate: ' +
-        'step 4 landed in objectui#6771 with this table non-empty, because everything left in it ' +
-        'is item-carried and outside the ruled family. ' +
-        'Re-point this block and say so in those terms; ⛔ do not delete the claim.'
-    ).toBeGreaterThan(0);
+      'a producer is BACK — shipped source emits the dialect again under objectui#9871\'s ' +
+        'EMISSION criterion, and it has not since objectui#9941. ⛔ Do not relax this to a ' +
+        'range: name the site and its carrier, and re-read the criterion before deciding ' +
+        'what a non-zero means here.'
+    ).toBe(0);
 
     // ⭐ The two SHAPES the census structurally cannot reach, asserted as
     // shapes rather than as paths: a producer arriving in a file nobody has
-    // named joins `producers` with no list to extend.
+    // named joins `producers` with no list to extend — that property is what
+    // this block holds, and it is unaffected by the occupancy being zero.
+    //
+    // ⚠️ B1's LIVE occupancy went to zero at objectui#9941; the three `tabs.tsx`
+    // `defaultProps` items were its last subjects. ⭐ The SHAPE is ⛔ not
+    // unreachable and this block is ⛔ not blind:
+    // `scripts/__tests__/body-dialect-producer-scan.test.ts`'s DIFFERENTIAL test
+    // plants `ITEM_PRODUCER` in a synthetic root and still asserts carrier
+    // ['item'], line 6, channel 'default-props' — it passed in the same run that
+    // first reported this red. That fixture is what makes standing down the
+    // live-tree control here safe rather than blind.
     expect(
       producers.filter((hit: { carrier: string }) => hit.carrier === 'item').length,
-      'no item-carried producer — the B1 shape (a `body` on an object with no `type`)'
-    ).toBeGreaterThan(0);
+      'an item-carried producer is BACK — the B1 shape (a `body` on an object with no `type`) ' +
+        'is live in shipped source again. ⛔ Not a licence to relax this; name the site.'
+    ).toBe(0);
     // ⚠️ B2 IS INVERTED BY THIS CARD, and the inversion is what the merge with
     // `main` actually found rather than a tidy-up. objectui#9871 asserted a
     // string-carried producer EXISTED, and its only subjects were the three VS
@@ -484,6 +509,9 @@ describe('the `body` consumers the ruling does not enumerate', () => {
     // `list`'s `items[].body`; `tabs` items and the `dashboard` widget key are
     // recorded on it by comment 5733850974, so this pointer resolves to a record
     // that actually carries the two shapes named here.
+    // ⚠️ VACUOUS SINCE objectui#9941 — the item-carried set is EMPTY, so this
+    // loop body never runs and asserts nothing. Kept, ⛔ not deleted: it is the
+    // disposition claim that fires the moment an item-carried producer returns.
     for (const hit of producers.filter((h: { carrier: string }) => h.carrier === 'item')) {
       expect(hit.disposition).toBe('unruled:item-carrier');
     }
diff --git a/scripts/__tests__/body-dialect-producer-scan.test.ts b/scripts/__tests__/body-dialect-producer-scan.test.ts
index 00ed54e894..bb4487aaf4 100644
--- a/scripts/__tests__/body-dialect-producer-scan.test.ts
+++ b/scripts/__tests__/body-dialect-producer-scan.test.ts
@@ -553,8 +553,47 @@ describe('the LIVE tree, read through the criterion', () => {
     // ⭐ Asserted as SHAPES, not as a count and not as a list of paths: a
     // producer arriving in a file nobody named appears in `producers` without
     // anyone extending anything, which is the whole point of this card.
-    expect(producers.filter((p: { carrier: string }) => p.carrier === 'item').length).toBeGreaterThan(0);
-    expect(files.size).toBeGreaterThan(0);
+    //
+    // ⭐ RE-POINTED at objectui#9941, ⛔ not deleted — this block's own former
+    // message asked for exactly that when the table emptied. What it now says,
+    // and it says ⛔ nothing beyond it:
+    //
+    //   1. the table is EMPTY **under objectui#9871's EMISSION criterion**
+    //      (C1 + C2). ⛔ That is not "the dialect has left the tree".
+    //   2. its last subject was the three `defaultProps` items in
+    //      `packages/components/src/renderers/layout/tabs.tsx`, respelled
+    //      `body:` -> `content:` under objectui#9941 because `TabItemSchema`
+    //      declares `content` required and declares no `body` — ⛔ a
+    //      CONFORMANCE fix, ⛔ not a dialect migration. The two are not one act.
+    //   3. ⛔ THIS IS NOT objectui#9590'S FINISH LINE AND THIS BLOCK DOES NOT
+    //      CLAIM IT IS. Measured across that same change: `failedC2` 313 -> 313
+    //      and `unclassified` 5 -> 5, both identical — 318 rows sit outside
+    //      what this criterion scores as emission and not one of them moved —
+    //      and objectui#9989 is a published doc still TEACHING the spelling.
+    //      ⭐ Emitted and taught are different verbs and this scan reads only
+    //      the first. Whether the finish line is reached is objectui#9590's to
+    //      judge on its own record.
+    //   4. ⭐ the instrument is ⛔ NOT blind, and that is what makes standing
+    //      down a live-tree control safe rather than blind: the DIFFERENTIAL
+    //      test at the top of this file plants `ITEM_PRODUCER` in a synthetic
+    //      root and still asserts carrier ['item'], line 6, channel
+    //      'default-props'. It passed in the same run that first reported these
+    //      reds. The SHAPE stays reachable; only its live occupancy is zero.
+    //
+    // ⛔ Do not relax either of these to a range — the file's own idiom for a
+    // shape that has emptied is `toBe(0)` plus a message naming what a non-zero
+    // means (see the string-carried assertion below).
+    expect(
+      producers.filter((p: { carrier: string }) => p.carrier === 'item').length,
+      'an item-carried producer is BACK — the B1 shape (a `body` on an object with no `type`) ' +
+        'is live in shipped source again, and it has not been since objectui#9941. ' +
+        '⛔ Not a licence to relax this: name the site and its carrier.',
+    ).toBe(0);
+    expect(
+      files.size,
+      'a producer FILE is back, so the empty-table reading above no longer holds. ' +
+        '⛔ Re-read that reading before changing anything here.',
+    ).toBe(0);
 
     // ⚠️ B2 INVERTED BY objectui#6771, and this is the SECOND copy of the claim
     // — the census block carries the first. ⭐ Neither file was in the merge's
@@ -580,6 +619,14 @@ describe('the LIVE tree, read through the criterion', () => {
       'lit control — no literal-carried hit passes C2, so the zero below would measure C2 ' +
         'rather than the emission channel it claims to measure',
     ).toBeGreaterThan(0);
+    // ⚠️ VACUOUS SINCE objectui#9941, and said out loud rather than left to be
+    // discovered by the next reader. `producers` is EMPTY, so every filter over
+    // it is 0 and this passes without searching for anything — the phantom-check
+    // shape. It is kept, ⛔ not deleted, because it is the claim that fires the
+    // moment any producer returns; until one does it measures nothing, and the
+    // two lit controls directly above are what keep B2's own sight under test in
+    // the meantime. ⇒ they are ⛔ not decoration now, they are the only live
+    // half of this paragraph.
     expect(
       producers.filter((p: { source: string }) => p.source !== 'code-key').length,
       'a string-carried producer is BACK — objectui#6771 step 5 migrated the last of them. ' +
@@ -601,15 +648,30 @@ describe('the LIVE tree, read through the criterion', () => {
     // `unknown-prop: <card> has no prop "body" — the child-list key is
     // "children"`, and a `tabs` ITEM carrying the same key draws ZERO
     // diagnostics. ⇒ step 4 refuses nothing the platform still ships.
-    // An empty table is objectui#9590's finish line, ⛔ not step 4's gate.
-    // ⛔ When this reds because the table emptied, that is still the HANDOFF —
-    // say so here in those terms rather than deleting the assertion.
+    //
+    // ⭐ RE-POINTED at objectui#9941, ⛔ not deleted — this assertion's own
+    // former message asked for that, in these terms. The table emptied when the
+    // three `tabs.tsx` `defaultProps` items were respelled to `content` to
+    // satisfy `TabItemSchema`, which declares `content` required and declares no
+    // `body`: ⛔ a conformance fix, ⛔ not a dialect migration.
+    //
+    // ⛔ An empty table is ⛔ NOT asserted here to be objectui#9590's finish
+    // line. The claim is scoped to what was measured — empty UNDER
+    // objectui#9871's EMISSION criterion. Across the same change `failedC2` held
+    // at 313 and `unclassified` at 5, so 318 rows outside what this criterion
+    // scores as emission did not move, and objectui#9989 is a published doc
+    // still TEACHING the spelling. Whether the finish line is reached is
+    // objectui#9590's to judge on its own record.
+    //
+    // ⭐ Safe rather than blind: the DIFFERENTIAL test at the top of this file
+    // plants `ITEM_PRODUCER` in a synthetic root and still asserts carrier
+    // ['item'], line 6, channel 'default-props'.
     expect(
       producers.length,
-      'the producer table is EMPTY — under this criterion nothing ships the dialect. That is ' +
-        'objectui#9590\'s finish line, ⛔ NOT objectui#6771 step 4\'s gate: step 4 landed with ' +
-        'this table non-empty and item-carried. Re-point this block, do not delete it.',
-    ).toBeGreaterThan(0);
+      'a producer is BACK — under objectui#9871\'s EMISSION criterion shipped source emits the ' +
+        'dialect again, and it has not since objectui#9941. ⛔ Do not relax this to a range: ' +
+        'name the site and its carrier, and re-read the criterion before deciding what it means.',
+    ).toBe(0);
     // ⏱ Explicit — one tree-wide scan over 5,136 files, measured 4.0s here; the
     // CI shard is at least 1.9x slower (objectui#9871's timeout reading), and
     // the default 15s leaves too little room on a loaded shard. Same 60s and the