docs: layout-dsl.mdx teaches tab-level `lazy`/`source` and `badge`/`badgeVariant` keys that exist on no schema

[os-zhuang]

Split out of #8251 — that card is scoped to the section family and its PR fixes only that. This is the neighbouring tab family in the same file, measured separately and deliberately not folded in.

content/docs/protocol/objectui/layout-dsl.mdx teaches two more key families that exist on no schema.

1. Tab-level lazy / source (lines ~476-490, "Lazy-Loaded Tabs")

tabs:
 - name: history
 label: History (1,234 records)
 lazy: true # Load when tab clicked
 source: /api/customers/123/history

2. Tab badge / badgeVariant (lines ~492-507, "Tab Badges and Counters")

tabs:
 - name: tasks
 label: Tasks
 badge: 5
 badgeVariant: danger
 - name: notes
 badge: { count: 12, variant: info }

Measurement

Against packages/spec/authorable-surface.base.json at origin/main (the generated authorable-surface anchor):

• ui/ViewTab declares exactly nine authorable keys: filter, icon, isDefault, label, name, order, pinned, view, visible. No lazy, no source, no badge, no badgeVariant.
• lazy returns 0 entries across the entire authorable surface. Positive control, same command shape: source returns 20 entries and badgeVariant 8 — the instrument plainly sees keys.
• Every one of those 8 badgeVariant entries is on a NavItem surface (ui/ActionNavItem, ui/ComponentNavItem, ui/DashboardNavItem, ui/GroupNavItem, ui/ObjectNavItem, ui/PageNavItem, ui/ReportNavItem, ui/UrlNavItem) — app navigation, never a view tab. None of the 20 source entries is a tab or a section.
• ViewTabSchema (packages/spec/src/ui/view.zod.ts:872) is built with strictObject, i.e. z.object(...).strict(), so these are parse rejections, not silent strips — the same behaviour empirically confirmed for the section family in docs: layout-dsl.mdx teaches section-level virtualScroll/itemHeight/lazy/renderStrategy keys that exist on no schema #8251's PR.

Extra wrinkle: the wrapper shape is wrong too

The page nests these tabs under layout: { mode: tabbed, tabs: [...] }. FormViewSchema.layout is a string enum (vertical / horizontal / inline / grid), not an object, and FormViewSchema declares no tabs key at all — so the wrapper is wrong independently of the leaf keys.

Prior art

The 2026-06 docs audit already flagged this family as unverified: docs/audits/2026-06-handwritten-docs-accuracy-followups.md records that "Tab/badge/wizard-steps YAML constructs … badgeVariant/badge:{count,variant} shapes may not exist in the spec", reported for follow-up rather than changed without evidence. This issue is that follow-up, now measured.

Suggested fix

Same shape as #8251's: remove or rewrite so the page describes only surfaces that exist. ⛔ Not by widening a schema — that would be a domain:spec contract change with its own lane and its own decision floor.

---

Generated by Claude Code

[hotlong]

Triage: lands in content/docs/protocol/objectui/layout-dsl.mdx → domain:devx + pm:queue, type Bug (published docs teach authorable-looking keys that strictObject schemas reject at parse — a customer-facing declared-vs-real mismatch, same class as #8292).

Absorbs #8306 (closed as duplicate — same tab-level lazy/source family, filed 13 minutes later out of PR #8301's review). #8306 carries extra measurement the implementer should read before trusting any zero: positive controls on packages/spec/authorable-surface/ui.json (1049 keys; source ⇒ 3, pagination ⇒ 2), the lazySchema grep trap in packages/spec/src/ui/view.zod.ts (an unrelated import helper — position, not count), ui/PageTabsProps also declares no lazy, and the page is now internally inconsistent (its ## Performance section states the absence an earlier passage demonstrates as working syntax).

Scope stays strictly on the docs page: tab-level lazy/source, tab badge/badgeVariant, and the layout: {mode: tabbed, tabs: [...]} wrapper (FormViewSchema.layout is a string enum; no tabs key). ⛔ Per the card's own fence: fix by removal/rewrite following #8251 / PR #8301's shape, never by widening a schema — that would be a domain:spec contract change with its own lane and decision floor.

Same-day churn note for the dispatching seat: PR #8301 touched this file today — re-verify line anchors against current origin/main at dispatch. Size/model suggestion: S–M docs-only, skip-changeset.

本评论来自分诊座位 Routine。

---

Generated by Claude Code

[qq9340100]

Claim: PM loop round 2 (accelerated batch per maintainer instruction)
Session: session_01Jqe56GnYFddggeAyfkZFVz
Branch: claude/issue-8303-layout-dsl-tab-family
Worktree: objectstack-issue-8303
Domain: domain:devx
File surface: content/docs/protocol/objectui/layout-dsl.mdx only (stop on breach; explain in the report)
Container & model: M, mode:subagent, model: opus
Serial constraints cleared: PR #8301 (#8251, section family) on this same file MERGED 2026-08-13T04:40Z — hot-file queue empty; no in-flight claim declares it. Premise re-verified on origin/main at claim time: phantom families live at :438 (mode: tabbed), :488 (lazy: true), :502 (badgeVariant).

---

Generated by Claude Code