fix(core,docs): bare citations that meant an objectstack card name its repository; the RETIRED_ACTION_KEYS.execute pointer among them (objectui#11016)

.changeset/11016-cross-repo-bare-citations.md

@@ -0,0 +1,34 @@
+---
+'@object-ui/core': patch
+'@object-ui/app-shell': patch
+---
+
+Citations of objectstack cards now name their repository (objectui#11016).
+
+A bare `#N` in this repository resolves to an objectui item. Where the sentence
+meant an objectstack card, a reader who followed the number landed on an
+unrelated objectui card or pull request, and nothing flagged it: both
+repositories answer for the number, so an earlier sweep that looked for dead
+numbers could not see this class.
+
+Two of those citations are runtime text, and they now name the card they mean:
+
+- `@object-ui/core`: the dev warning for an action that still carries the
+  retired `execute` key (`RETIRED_ACTION_KEYS.execute`) points at
+  objectstack-ai/objectstack#3855, the card that retired the alias. The rest of
+  the message, including "rename the key to `target`", is unchanged.
+- `@object-ui/app-shell`: the Studio skill form's help text for
+  `triggerConditions` points at objectstack-ai/objectstack#1878 and
+  objectstack-ai/objectstack#1895.
+
+The other sites are code comments and doc comments in the non-test sources,
+requalified the same way: a bare number whose sentence meant
+an objectstack card now reads `objectstack-ai/objectstack#N`. Numbers that mean
+an objectui card stay bare. Where a comment quotes objectstack's own text, the
+quote keeps its words and the card name sits in square brackets, the editorial
+mark for a substituted word.
+
+Pending entries in this release that carry such a number get a dated note
+naming the card, and their frontmatter is unchanged.
+
+Nothing else moves: no key, accept set, refusal decision or code path.

.changeset/4664-related-list-declared-filter.md

@@ -16,6 +16,11 @@ related lists answered wider than the metadata asked, and the driving scenario 
 soft-deleted child rows on auto-derived record pages — had no way to be
 expressed at all.
 
+⚠️ **Dated note, 2026-09-29 — the pull request beside objectstack#8704 is objectstack's — objectui#11016.** The spec key's
+pull request is named above by a bare number, which in this repository resolves
+to an unrelated objectui pull request. It is objectstack-ai/objectstack#8955, the
+spec change that declared the field-level `relatedListFilter`. The text above is kept as the reading of this change.
+
 What ships:
 
 - `deriveRelatedLists` reads the key off the FK and carries it on the derived

.changeset/6956-listview-export-options-spec-mirror.md

@@ -32,6 +32,11 @@ intact. The mirror now IS that schema object, so the four verdicts are the
 spec's by construction; `export-options-spec-parity.test.ts` pins the identity,
 the four verdicts, and the survival of `streaming` through a parse.
 
+⚠️ **Dated note, 2026-09-29 — the card behind the mirror's old shape is objectstack's — objectui#11016.** The shape the
+mirror used to declare is dated above by a bare number, which in this repository
+resolves to an unrelated objectui item. It is objectstack-ai/objectstack#8010, the
+card the first paragraph cites as objectstack#8010. The text above is kept as the reading of this change.
+
 **The TS face follows.** `ListViewSchema['exportOptions']` is now the spec's
 INPUT type: `ListViewExportFormat[] | ListViewExportOptions` — the bare array
 stays admissible on input because nothing on the render path parses, and the

.changeset/7122-spec-17-3-0-pin-repointing.md

@@ -47,3 +47,10 @@ released by this change.
   refusal's SHAPE. The `LocationField` fence explicitly re-derives why its
   key-level assertions stay: no parse stands on the emit path, so strictness
   changed the consequence of a spread regression, not the widget's exposure.
+
+⚠️ **Dated note, 2026-09-29 — the ruling number in the first two bullets is objectstack's — objectui#11016.** The first
+two bullets name the `submitBehavior.url` ruling by a bare number, once as the
+family's name and once inside the quoted 17.2.0 spelling. In this repository that
+number resolves to an unrelated objectui item; it means
+objectstack-ai/objectstack#7496, the spec card that ruled the semantics. The quoted
+spelling is the artifact's own and stays verbatim. The text above is kept as the reading of this change.

.changeset/9137-dataset-drill-conjoin-widget-filter.md

@@ -38,6 +38,12 @@ multi-condition dataset drill — including paths pinned by objectui#9085,
 objectui#4056, objectstack#5473 and #1752 — for no change in the rows selected
 or the `filter[...]` params emitted.
 
+⚠️ **Dated note, 2026-09-29 — the last pinned path in the list above is objectstack's — objectui#11016.** The list of
+pinned paths above ends with a bare number after objectstack#5473, which in this
+repository resolves to an unrelated objectui pull request. It is
+objectstack-ai/objectstack#1752, the card that made a time-bucketed date dimension
+drill by range rather than by equality. The text above is kept as the reading of this change.
+
 **The failure direction, corrected by measurement.** This was filed and graded
 as a silent fail-OPEN superset, on the premise that an index key is one "nothing
 reads as a condition". It is not: a bare array is not a legal equality comparand

.changeset/dashboard-aria-member-retired-5830.md

@@ -9,6 +9,11 @@ claimed alignment with `@objectstack/spec AriaPropsSchema`, but the spec removed
 a tombstone that refuses any value and tells authors to delete the key — and no
 dashboard renderer ever read `schema.aria` (objectui#5830).
 
+⚠️ **Dated note, 2026-09-29 — the audit close-out named above is objectstack's — objectui#11016.** The audit close-out
+is cited above by a bare number, which in this repository resolves to an unrelated
+objectui card. It is objectstack-ai/objectstack#3896, whose audit close-out is where
+the spec retired `dashboard.aria`. The text above is kept as the reading of this change.
+
 What an author loses is the **type-level suggestion** only: the key was already
 refused at parse (the Zod twin inherits the spec tombstone by reference), and
 `BaseSchema`'s index signature means an existing `aria:` line still compiles.

.changeset/designer-publish-package-binding.md

@@ -27,3 +27,8 @@ narrowing at the runtime publish gate fire on an HTTP-driven promotion at all. T
 narrowing has a second, independent gate this does not touch — objects carrying no
 `_packageId` provenance are kept unconditionally — so on a tenant-authored overlay
 corpus stating the package still narrows nothing.
+
+⚠️ **Dated note, 2026-09-29 — the publish-gate narrowing is an objectstack card — objectui#11016.** The last paragraph
+names the package-closure narrowing by a bare number, which in this repository
+resolves to an unrelated objectui pull request. It is
+objectstack-ai/objectstack#9612, the runtime publish gate's rule snapshot. The text above is kept as the reading of this change.

.changeset/gantt-navconfig-comment-once.md

@@ -12,4 +12,10 @@
   `width` default supplies the identical `min(960px, 60vw)`. The resolved
   overlay width is unchanged on every viewport, and is now pinned by a test.
 
+⚠️ **Dated note, 2026-09-29 — the card in the quoted deprecation tag is objectstack's — objectui#11016.** The second
+bullet quotes the spec's deprecation tag for `width`, whose bare number resolves
+in this repository to an unrelated objectui item. It is
+objectstack-ai/objectstack#2578, the spec card behind the `size` bucket that
+replaces `width`. The text above is kept as the reading of this change.
+
 No published behaviour changes.

.changeset/kanban-calendar-nav-width-converge.md

@@ -14,6 +14,12 @@ calendars take. With both omitted, `resolveOverlayWidth` returns `undefined` and
 `RecordDetailDrawer`'s own `width` default supplies the identical
 `min(960px, 60vw)`.
 
+⚠️ **Dated note, 2026-09-29 — the card in the quoted deprecation tag is objectstack's — objectui#11016.** The paragraph
+above quotes the spec's deprecation tag for `width`, whose bare number resolves
+in this repository to an unrelated objectui item. It is
+objectstack-ai/objectstack#2578, the spec card behind the `size` bucket that
+replaces `width`. The text above is kept as the reading of this change.
+
 The resolved overlay width is therefore unchanged on every viewport, for both a
 board/calendar that declares no `navigation` and one that authors
 `navigation.width` — each is now pinned by a test. The three renderers agree

.changeset/render-publish-advisory-findings-5026.md

@@ -10,6 +10,12 @@ objectui#4133 / PR #4236 wired the gate's advisories to the save door and record
 
 The second reason has expired. `PublishMetaItemResponseSchema` now declares the same optional, omitted-when-empty `advisories` key that `SaveMetaItemResponseSchema` has carried since #4717, and `publishMetaItem` populates it. Measured against the installed `@objectstack/spec` (17.2.0) rather than inferred from the version number: the key survives a `safeParse`, a half-shaped finding is rejected, and a clean publish omits the key entirely. That reading is now a test rather than a note, so a spec drift fails CI instead of silently re-muting the door.
 
+⚠️ **Dated note, 2026-09-29 — the save response's advisories key came from an objectstack card — objectui#11016.** The
+paragraph above dates the save response's `advisories` key by a bare number,
+which in this repository resolves to an unrelated objectui item. It is
+objectstack-ai/objectstack#4717, the card that put the runtime authoring gate's
+advisory findings on the response. The text above is kept as the reading of this change.
+
 `MetadataClient.publish` and `MetadataClient.publishDraft` — the two methods over the single-item publish route `POST /meta/:type/:name/publish` — now report through the **same** sink, the same event and the same renderer the save door already used. No new UI shape: same warning tier, same 10s duration, same per-finding `rule` + `message` + `hint` formatting, findings still rendered verbatim as server prose. The wiring lands in the data layer rather than at the call sites, so `ResourceEditPage`'s Publish button and the runtime `RuntimeDraftBar` promotion (ObjectView / ReportView / DashboardView) are covered by one change, as are future ones.
 
 One thing had to differ, and it is the frame's verb. Save and Publish are two different buttons in this product, so a toast that says "Saved" after a Publish tells the author their change is still a draft — the opposite of what happened. `MetadataSaveAdvisoryEvent` therefore gains a required `door: 'save' | 'publish'` and the renderer picks `console.publishAdvisoryTitle` (added to all ten locale packs) accordingly. `door` exists because `mode` cannot answer this: a direct active save and a draft promotion both report `mode: 'publish'`, since both land the body in the active overlay. It is required rather than optional so a future third door cannot be wired without saying which one it is, and the renderer branches on it through an exhaustive switch with a `never` check, so adding a third member is a compile error rather than a silently wrong verb.

.changeset/violet-donkeys-brake.md

@@ -6,4 +6,10 @@ PackageFormDialog: a producer-marked `error.userMessage` now outranks the locali
 
 Creating or saving a package refused with **409** (id taken) or **403** (no `manage_metadata` capability) answered with a localized constant and discarded the envelope's prose — including a `userMessage` the producer had marked, at throw time, as addressed to the person reading the screen (ADR-0112 / objectui#3821). Those are the two refusals an author meets most on this surface, so the one channel written for them was dropped exactly where it mattered.
 
+⚠️ **Dated note, 2026-09-29 — the card cited beside ADR-0112 is objectstack's — objectui#11016.** The
+paragraph above cites the generic-substitution card under this repository's name,
+which here resolves to an unrelated objectui pull request (the check-i18n-en-drift
+script). It is objectstack-ai/objectstack#3821, the sharing-rule dialog card, whose
+objectui repair (objectui#2920) introduced the generic substitution. The text above is kept as the reading of this change.
+
 Both arms now render a marked sentence when the envelope carries one, and keep the localized constant as the fallback for an **unmarked** body — which is every refusal the package doors emit today, so the settled localized posture for a withheld capability (objectstack#8270) is unchanged. This also removes a divergence that depended on the transport: the same marked body used to read one way when an HTTP status survived the wire and another way when it did not.

packages/app-shell/src/chrome/CommandPalette.tsx

@@ -133,7 +133,7 @@ export function CommandPalette({ apps, activeApp, objects, onAppChange, dataSour
   }, [objects]);
 
   // Group the (server-ranked) record hits by object so the palette lists them
-  // under per-object headings — issue #3371 asks for record hits "grouped by
+  // under per-object headings — issue objectstack-ai/objectstack#3371 asks for record hits "grouped by
   // object". The object with the top-ranked hit leads (first-seen order), and
   // within each group the server's relevance order is preserved.
   const recordGroups = useMemo(() => {
@@ -216,7 +216,7 @@ export function CommandPalette({ apps, activeApp, objects, onAppChange, dataSour
         )}
 
         {/* Record search hits from the platform's global search
-            (/api/v1/search), grouped by object (issue #3371). The searching
+            (/api/v1/search), grouped by object (issue objectstack-ai/objectstack#3371). The searching
             pulse rides the first group's heading so it stays visible while a
             refined query is in flight over an existing result set. */}
         {recordGroups.map((group, groupIndex) => {

packages/app-shell/src/console/AppContent.tsx

@@ -210,7 +210,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
   // skips hidden apps to avoid landing the user on a personal-settings
   // app by default.
   //
-  // ⛔ `hidden` only — never `_unpublished` (objectstack#6955 / #4829 A1). The
+  // ⛔ `hidden` only — never `_unpublished` (objectstack#6955 / objectstack-ai/objectstack#4829 A1). The
   // publish gate is enforced server-side; duplicating it here would strand a
   // builder on their own unpublished app.
   const activeApps = apps.filter((a: any) => a.active !== false);
@@ -395,7 +395,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
     }
   }, [activeApp?.name, appName, location.pathname, navigate]);
 
-  // #2604 — the create/edit overlay is URL-driven (`?form=new` / `?form=<id>`),
+  // objectstack-ai/objectstack#2604 — the create/edit overlay is URL-driven (`?form=new` / `?form=<id>`),
   // not component state: the record form is a TASK overlay over the origin
   // route, and putting its open-state in the URL makes browser Back close the
   // overlay (returning to the intact origin) instead of abandoning the route
@@ -405,7 +405,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
   // ahead in history.
   const [searchParams, setSearchParams] = useSearchParams();
   const recordFormParam = searchParams.get(RECORD_FORM_PARAM);
-  // #2604 D3 — child-task extension of the record-form URL contract:
+  // objectstack-ai/objectstack#2604 D3 — child-task extension of the record-form URL contract:
   // `formObject` names the object the form edits when it is NOT the route's
   // object (a subtable child opened over its parent's detail); `formLink`
   // ("field:id") pre-links the parent on create. Keeping the whole task in
@@ -563,7 +563,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
   const currentObjectDef = allObjects.find((o: any) => o.name === objectNameFromPath);
 
   // The object the record-form overlay edits: the route's object by default,
-  // or the `formObject` child override (#2604 D3 — subtable child task opened
+  // or the `formObject` child override (objectstack-ai/objectstack#2604 D3 — subtable child task opened
   // over its parent's detail).
   const formObjectDef = formObjectParam
     ? allObjects.find((o: any) => o.name === formObjectParam)
@@ -586,15 +586,15 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
     executeAction({ type: 'dialog_cancel' });
   }, [executeAction]);
 
-  // #2604 save invariant — *edit never moves you; create takes you to the
+  // objectstack-ai/objectstack#2604 save invariant — *edit never moves you; create takes you to the
   // record you made.* Edit save: the origin route is untouched (crud_success
   // bumps refreshKey → origin refetches in place). Create save: land on the
   // new record's detail, on ITS derived surface — a light object's detail is
   // the drawer over the still-intact list; a heavy one is the detail route.
   // `replace: true` swaps out the transient `?form=…` entry so Back returns
   // to the pre-create origin.
   const handleRecordFormSuccess = useCallback(async (saved: any) => {
-    // Child task (#2604 D3): the parent detail must stay EXACTLY as it was —
+    // Child task (objectstack-ai/objectstack#2604 D3): the parent detail must stay EXACTLY as it was —
     // active tab, scroll, everything. So do NOT go through crud_success;
     // the child's open related lists refetch on their own via the
     // invalidation bus (#2269): the dataSource write already emitted a
@@ -1025,7 +1025,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
           no organization), and widening it for filter resolution would couple
           two unrelated contracts. Renderer packages deliberately do not depend
           on @object-ui/auth, so the shell supplies the values (framework
-          #3574). */}
+          objectstack-ai/objectstack#3574). */}
       <FilterScopeProvider
         currentUserId={user?.id ?? null}
         currentOrgId={activeOrganization?.id ?? null}
@@ -1146,10 +1146,10 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
                 objectName: formObjectDef.name,
                 mode: editingRecord ? 'edit' : 'create',
                 recordId: editingRecord?.id,
-                // Child create task (#2604 D3): pre-link the parent from the
+                // Child create task (objectstack-ai/objectstack#2604 D3): pre-link the parent from the
                 // `formLink` URL param (refresh-safe — the link survives).
                 ...(formLinkValues && !editingRecord ? { initialValues: formLinkValues } : {}),
-                // #2604 D1: create/edit follow the flow-surface derivation —
+                // objectstack-ai/objectstack#2604 D1: create/edit follow the flow-surface derivation —
                 // field-heavy → full-screen modal (the same big canvas the
                 // detail page gets, with overlay return semantics); light
                 // objects keep the existing auto-sized modal (ModalForm
@@ -1170,7 +1170,7 @@ export function AppContent({ extraRoutes, extraRoutesNoApp }: AppContentProps =
                 // When the view declares sections they drive the modal layout and
                 // win over the flat `fields` list below; otherwise this resolves to
                 // {} and `fields` (every field, raw schema order) is used as before.
-                // `formType` stays 'modal' (the container). (#1890 / ADR-0050.)
+                // `formType` stays 'modal' (the container). (objectstack-ai/objectstack#1890 / ADR-0050.)
                 ...resolveFormViewLayout(formObjectDef as any),
                 title: editingRecord
                   ? t('form.editTitle', { object: objectLabel(formObjectDef as any) })
@@ -1255,7 +1255,7 @@ function buildItemRoute(item: any, ctx?: NavTemplateContext): string {
  * the ROOT lands on is a separate question, decided by `isDefault`.
  *
  * This used to honour an explicit `app.homePageId`. Spec 17.0.0 retired that
- * key (objectstack#4667, premise corrected in #4709 / objectui#3287): it was
+ * key (objectstack#4667, premise corrected in objectstack-ai/objectstack#4709 / objectui#3287): it was
  * an ID cross-reference with no referential integrity, so a dangling id fell
  * back to the first item *silently*. If the capability returns, the correct
  * encoding is a flag on the navigation item itself (`navigation[].landing`),