types: drop `export type * as Cloud from '@objectstack/spec/cloud'` before the upstream subpath is removed (step 2 of 3 of objectstack#16325)

[os-zhuang]

Step 2 of 3 of the objectstack#16325 ruling (director seat, decision batch #62, 2026-09-07, option B): the cloud control-plane contracts leave @objectstack/spec; the ./cloud subpath export is deleted upstream once cloud and objectui no longer import it.

Blocked-by: #7122

Scope

• packages/types/src/index.ts:1106 — export type * as Cloud from '@objectstack/spec/cloud'; is the only real dependency in this repo (measured at 0558e0f78; every other hit is a comment saying the type was deliberately re-declared inline). Remove the line.
• If any consumer of the Cloud namespace exists in this repo or in hotcrm, re-point it to the inline declarations already used by marketplaceApi.ts:85, usePackageL10n.ts:6, PackagesPage.tsx:75; record the count here (expected 0, with a firing control).
• No behaviour change; changeset states the removed type namespace (public @object-ui/types surface, so Clause-② yes — landing PR carries needs:contract-review).

Acceptance

• zero imports of @objectstack/spec/cloud in objectui (control on the same grep)
• @object-ui/types surface pin regenerated with the repo tooling
• objectstack#16325 notified when landed

Refs objectstack#16325, objectstack#12036.

[hotlong]

Yielded on a hot-file collision, not dispatched — pm:dispatched rolled back to pm:blocked

domain:spec cross-repo lane, direct maintainer dispatch session session_f95e3874-e532-4748-a921-044aa2752a2b,
readings 2026-09-07T06:0xZ. objectui origin/main cdbc852e9 (the card's reading was 0558e0f78).

Maintainer authorisation for the dispatch attempt, verbatim (⛔ quoted, not translated): 「这个作为专题卡，你直接派发」.

Why this card was NOT dispatched

The pre-dispatch targeted in-flight check found a declared file-surface intersection. objectui#7122 is
in flight — assignee os-justin, session session_01BAZFhALsQsGqxui8sNqM8s, branch
claude/issue-7122-spec-pin-17-3-0 (continuing PR #7685) — and its claim comment declares:

File surface: pnpm-lock.yaml · packages/types/** (src, zod mirrors, __tests__) · …

packages/types/** covers packages/types/src/index.ts, which is this card's entire target. Same-file work
serialises across rounds; intersect ⇒ yield. objectui#7635 rides the same branch as item 2 of that chain.

Blocked-by: objectstack-ai/objectui#7122 added to the body so the unlock scan returns this card when that
chain lands, rather than leaving the constraint as prose nobody re-reads.

⚠️ This lengthens the objectstack#16325 chain by one hop: objectstack#16325 ← this card ← objectui#7122.
Reported to the maintainer in the same session.

Premise correction for whoever picks this up

The card's packages/types/src/index.ts:1106 is stale — objectui main moved 0558e0f78 → cdbc852e9.
Re-measured at cdbc852e9:

$ git show origin/main:packages/types/src/index.ts | grep -nE "export type \* as Cloud|@objectstack/spec/cloud"
1116:export type * as Cloud from '@objectstack/spec/cloud';

Line 1116, single occurrence. It sits inside a family of sibling namespace re-exports (UI, System,
AI, API, Automation, …); control on the same file: 15 export type * as lines total, so removing the
Cloud one leaves a deliberate gap that the changeset must name.

Consumer count re-confirmed 0, with a firing control:

$ git grep -nE "\bCloud\.[A-Z]|from .*types.*\{[^}]*\bCloud\b" origin/main -- 'packages/**/*.ts' 'packages/**/*.tsx' \
    | grep -v packages/types/src/index.ts
(no output)
$ git show origin/main:packages/types/src/index.ts | grep -c "^export type \* as"
15

⇒ the zero is a real absence, not a dead pattern.

State

pm:dispatched + assignee were set as step ① of the claim pair and are rolled back in the same session
— assignee cleared, label replaced with pm:blocked. ⛔ No half of the pair is left standing: no Claim:
comment was ever posted for this card, and no agent was dispatched.

[hotlong]

Claim: PM direct maintainer dispatch (no numbered round — direct-dispatch channel); serial hold cleared
Session: session_f95e3874-e532-4748-a921-044aa2752a2b
Branch: claude/issue-8225-drop-cloud-namespace-reexport
Worktree: objectui-issue-8225
Domain: domain:spec (cross-repo spec lane)
File surface: packages/types/src/index.ts (one export line) · the @object-ui/types surface pin regenerated by this repo's own tooling · .changeset/ (stop on breach; explain in the report)
Container & model: S, mode:subagent, model: claude-fable-5-1 (= CONTRACT_REVIEW_TIER; mandated by clause ② below — removing a published type namespace from @object-ui/types — not by a path glob)
Clause-②: yes
Serial constraints cleared: the hold recorded in comment 5565861943 is released on measurement, not on assumption — see below. objectui#7122 closed/completed, its chain PR #7685 merged 2026-09-07T07:48:20Z. objectui#7635 is still open and pm:dispatched to os-justin on that same chain, but its real sites are disjoint from this card's single line (measured below). No other in-flight objectui card declares packages/types/src/index.ts.

---

Maintainer statement that prompted the re-check, verbatim (⛔ quoted, not translated): 「7122 已经合并了」.
Standing dispatch authorisation for this chain, verbatim: 「这个作为专题卡，你直接派发」.

⚠️ Dispatched under the direct-dispatch channel while the domain:spec seat post (objectstack#6017) has an
incumbent. ⛔ No seat was seized; the authorisation above is what routes it.

Why the hold is released — and why #7635 being open does not re-impose it

The hold in 5565861943 was a hot-file collision with objectui#7122's declared surface, packages/types/**.
#7122 is closed, but its chain item #7635 is still open and dispatched on the same branch, so "the chain
landed" is not by itself enough. Measured on origin/main 411a132c0:

#7635's declared sites are three packages/types/src/__tests__/* files plus the objectui-side extension members
SelectOptionSchema.description / FieldSchema.rows / ObjectSchema.editMode, and two sites added by later
cross-lane comments (packages/app-shell/src/__tests__/spec-symbol-parity.test.ts,
apps/console/src/__tests__/registry-inputs-spec-parity.test.ts). Do any land in index.ts?

$ git show origin/main:packages/types/src/index.ts | grep -c SelectOptionSchema   → 0
$ …                                                | grep -c FieldSchema          → 0
$ …                                                | grep -c editMode             → 0
$ …                                                | grep -n  ObjectSchema
515:  ObjectSchemaMetadata,
516:  ObjectSchemaClientExtensions,

⇒ The only ObjectSchema hits are different symbols sharing a name prefix, not the ObjectSchema.editMode
member. That member's real declaration is elsewhere:

$ git grep -ln editMode origin/main -- 'packages/types/src/**'
packages/types/src/__tests__/object-schema-metadata-spec-derivation.test.ts
packages/types/src/field-types.ts

⇒ Intersection with this card's single line is empty, and provably so. ⛔ Nothing in #7635's work touches
packages/types/src/index.ts.

The Blocked-by: objectstack-ai/objectui#7122 line stays in the body as the record of why this card waited — a
closed target reads as satisfied; ⛔ deleting satisfied edges erases the history.

Premise re-check at dispatch time (⛔ not inherited — the card's own number is stale twice over)

objectui origin/main 411a132c0. The card body says packages/types/src/index.ts:1106; comment 5565861943
corrected it to 1116. Both are stale. Now:

$ git show origin/main:packages/types/src/index.ts | grep -nE "export type \* as Cloud"
1120:export type * as Cloud from '@objectstack/spec/cloud';

Line 1120, single occurrence. Consumer count re-confirmed 0 across packages/** and apps/**
(\bCloud\.[A-Z], excluding the declaring file), against a control of 15 export type * as lines in the same
file ⇒ the zero is a real absence, not a dead pattern. ⛔ Whoever works this counts on the tree in front of them;
the line has moved three times in one day.