docs(audits): flag dead-surface plan as invalidated (objectui-blind); cross-repo correction

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: os-zhuang

.changeset/reaudit-crossrepo.md

@@ -0,0 +1,4 @@
+---
+---
+
+fix(spec): cross-repo liveness re-audit — reclassify 16 props dead→live (objectui authoring UI) and flag the dead-surface plan as invalidated. The 2026-06 audits missed objectui's metadata-admin authoring layer; verified each correction by reading the consumer code in both repos. Repo-internal tooling/docs; no package impact.

docs/audits/2026-06-dead-surface-disposition-plan.md

@@ -1,5 +1,9 @@
 # Dead-surface disposition plan (spec liveness)
 
+> ## ⚠️ CORRECTION (2026-06-16) — this plan is substantially invalidated; do NOT prune from it
+> The "dead" verdicts were seeded from audits that **only checked the framework runtime + end-user renderers and MISSED objectui's metadata-admin authoring UI** (`../objectui/packages/app-shell/src/views/metadata-admin/`, which renders/edits nearly every authorable prop). A cross-repo re-audit (reading the actual consumer code in BOTH repos) found **~25 props this plan called "dead" are actually objectui-live** — pruning them would break the authoring UI / detail rendering. Reclassified `dead→live` in the ledgers: field `externalId`/`currencyConfig`/`searchable` + the 6 master-detail overrides; tool `category`/`requiresConfirmation`/`active`/`builtIn`; skill `triggerPhrases`/`permissions`; action `execute`/`shortcut`/`bulkEnabled`/`disabled`; flow `status`/`active`/`nodes.boundaryConfig`; agent `knowledge`; role `parent`; permission `isProfile`.
+> **Only ~42 props are confirmed dead in BOTH repos** (the genuine prune candidates — dominated by the 24 object-level props, which also hit `NoExcessObjectKeys` TS-strict cross-package coupling). Merged prunes #1945/#1946 were re-verified against objectui and are safe. **Any further pruning MUST verify each prop against framework AND objectui first.** The sections below are kept for history but their per-prop "prune" verdicts are unreliable.
+
 **Date**: 2026-06-16 · **Input**: the spec-liveness gate (`packages/spec/liveness/`) over 10 governed metadata types. **Builds on**: [ADR-0049](../adr/0049-no-unenforced-security-properties.md) (security enforce-or-remove). **Purpose**: turn the gate's measurement into an executable decision — for each of the **89 `dead` properties**, recommend **enforce / wire-rename / prune / mark-experimental / defer / objectui** with rationale, cost and risk.
 
 > The gate already proves these properties have no framework runtime consumer. This document decides what to *do* about each. It makes recommendations; the team owns execution (especially spec removals and runtime enforcement).