SUPERTRACK E. Goal: readiness never uses regex, function names, or object-existence as capability proof.

Keyset/resume/refresh/perf are now a DATA contract, not a name lookup. New table qt001_capability_contract holds four capabilities (keyset_pagination, resume_watermark, automatic_refresh, representative_perf_gate), each satisfied=false by default with a CHECK that a row may only be satisfied if it carries a proof_object and a non-NONE proof_kind. v_qt001_scale_blocking_guard_v3 reports capabilities_satisfied 0 of 4, scale_not_safe true, scale_safe false. This replaces the old proname-regex keyset gate.

v_qt001_apply_readiness_dashboard_v6 has eleven gates, each derived from a data-driven guard or a count — never from object names or source text. SSOT gates: rule_governance_enforced (governance guard), structural_hardcode_clean (hardcode v4), plan_fingerprint_v3_complete (all sensitivity tests pass AND fingerprint scope_count == snapshot count), exact_signoff_logic_sound (signoff v3 guard), parity_not_authority (pure data: zero GRANT rules on parity/divergence), public_execute_blocking (no-go guard), directus_dml_blocking (no-go guard). APPLY gates: exact_plan_bound_signoff_present, tier_permits_apply, owner_execute_permit_valid, scale_capability_satisfied. v_qt001_apply_readiness_guard_v6 aggregates: overall_ready false, SSOT 5/7 green (the two red SSOT gates are public_execute_blocking and directus_dml_blocking), APPLY 0/4, computed_status BLOCKED.

Negative tests: v_qt001_readiness_v6_negative_tests — five tests, all pass: unsatisfied capability blocks scale; no exact signoff blocks apply; public/directus open blocks overall; readiness currently BLOCKED; and a diagnostic confirming the v6 dashboard definition contains no proname / pg_proc / pg_get_functiondef (the source-text-absence check lives here as a test, NOT in the authority path). The readiness authority chain is purely data/behavioral.