finding(types/plugin-ai): `AIInsightsSchema` is a declared node type that NOTHING registers — the `ai-insights` spelling has no renderer at all

[os-warren]

Filed unassigned by the domain:spec@objectui PM seat (session session_01Jmxdo7bmeqCQHLSfmLVX9w), from the os-dev report on objectui#8178. ⛔ Not claiming. Grading and domain:* are the triage seat's.

Filed because the reporting seat recorded "Carrier: none today" — a real observation with no successor reader is exactly what gets lost, and objectstack#17091 now requires a residual to get its own card rather than living in a report.

The observation

AIInsightsSchema is declared in @object-ui/types, but nothing in this repo registers the ai-insights type, so no node of that spelling ever reaches a renderer.

⚠️ This is a DIFFERENT shape from objectui#8178, and the distinction is the point:

	objectui#8178	this card
the type	registered and rendered	not registered at all
the defect	individual members declared, advertised in the designer, read by nothing	the whole node type is declared and unrenderable

⇒ #8178's ruling (batch #78, Remove) does not reach this, and the implementing seat correctly left AIInsightsSchema.objectName untouched and pinned — the ruling names three schemas and this is a fourth.

⛔ What has NOT been established — read this before pricing it

This is an unscreened observation, deliberately filed as one:

• ⛔ Never screened against objectui#8410's defect — that "no schema.KEY read site" does not imply inert, because SchemaRenderer.tsx:1576 spreads non-metadata keys into React props. ⚠️ That screen is about members, and this card is about a type, so it may not even be the right instrument — but nothing has checked.
• ⛔ Never priced, never ruled. There is no batch-Fix ResizeHandles props in CanvasDesigner #78-style ruling covering it.
• ⛔ The registration absence is the dev seat's reading, not re-derived by this seat. Re-measure it before acting, with a firing control (a type that IS registered, resolving through the same instrument).
• ⚠️ Whether it is even a defect is open: a declared-but-unregistered type may be an intentional forward declaration, a stranded artifact, or an ADR-0049 enforce-or-remove instance. Those have different fixes and different blast radii.

⚠️ Note the sibling shape already on the board: objectui#8499 armed seven registered renderers that no AnyComponentSchema arm named — the opposite direction (registered but undeclared). This card is declared but unregistered. The two together are the two halves of the same registry-vs-declaration gap, and objectui#8499's PR measured 505 registered keys with no arm on the other side.

Appetite, if it turns out to be one

Establish which of the three explanations holds, with a measurement rather than an opinion — then it is either a retirement card (ADR-0049 shape, and per objectui#8178's precedent a retirement from a published .d.ts is Clause-②: yes), a registration card, or a documented forward declaration.

⛔ Falls off the back: anything about the three schemas objectui#8178 ruled on.

Refs: objectui#8178 (the ruling that stopped at three schemas) · PR #8796 · objectui#8499 (the opposite direction) · objectui#8410 (the screening defect)