07 - No-Disposition Guard Scope Fix

What Codex required

G-LEGACY-NO-DISPOSITION-AUTHORITY must:

• scan ACTIVE_AUTHORITY load-bearing docs/sections;
• separately report old terms in SUPERSEDED_NON_AUTHORITY as history, not fail;
• fail if old disposition/STUB terms appear in ACTIVE_AUTHORITY;
• fail if any future package references a SUPERSEDED_NON_AUTHORITY instruction;
• fail if the boundary itself is absent, ambiguous, or Directus-editable.
• depend on ACTIVE_AUTHORITY scope.

What T1 patched (blueprint doc 06)

G-LEGACY-NO-DISPOSITION-AUTHORITY input is changed from "the whole blueprint" to the ACTIVE_AUTHORITY load-bearing docs/sections (per the doc 00 §Active-authority boundary registry; depends on G-ACTIVE-AUTHORITY-SCOPE). Its expected output now states: over ACTIVE_AUTHORITY only — no disposition enum / LEGACY_* row / disposition column-family / classifier-CASE / STUB-body path / DO_NOT_TOUCH subtraction / relkind-prokind-name-owner-pattern-label policy. Fenced SUPERSEDED_NON_AUTHORITY history is scanned separately and reported, never failed or consumed. Failure now triggers if a removed construct re-appears in ACTIVE_AUTHORITY, or a future package/guard/seal consumes a SUPERSEDED_NON_AUTHORITY instruction.

Two supporting guards back it:

• G-ACTIVE-AUTHORITY-SCOPE (the boundary is present/unambiguous/not-Directus-editable — its hard dependency; if the boundary is broken, this guard fails before the no-disposition scan can give a false green).
• G-NO-SUPERSEDED-CONSUMPTION (no package/guard/seal/order/authoring step consumes a superseded section).

Guard-quality rule 8 binds every text-scanning guard to the ACTIVE_AUTHORITY scope and the report-not-fail behavior for fenced history.

Why the consistency defect (Codex A/06) is resolved

Before: old disposition/STUB instructions survived in checkpoint/final-history sections that the guard could not distinguish from current normative text, so the guard could not both reject those constructs and pass. After: those sections are fenced as SUPERSEDED_NON_AUTHORITY (doc 00/12 + checkpoint), the guard scopes to ACTIVE_AUTHORITY, and the few residual stale phrases in load-bearing docs (the doc 00 self-verdict; doc 00 central-finding "freeze/deprecate"; the checkpoint central finding) were either fenced or reworded to the uniform end-state. The guard now passes on ACTIVE_AUTHORITY and reports the fenced history as audit trail.

Self-check

PASS. The guard fails on old terms in ACTIVE_AUTHORITY, reports (does not fail on) fenced history, fails on superseded consumption, and depends on G-ACTIVE-AUTHORITY-SCOPE.