onto-mcp 0.4.12 → 0.4.13

package/.onto/processes/reconstruct/obligation-coverage-ledger.yaml

@@ -0,0 +1,842 @@
+# Obligation-coverage PENDING ledger (G(a) / INV-OBLIGATION-COVERAGE-1).
+# Generated; the parked backlog of active (validator_id, obligation_id) pairs not yet RECORDED
+# (no asserted_obligation_ids instrumentation in this slice). The ratchet proves this legacy set
+# is monotone non-increasing.
+#
+# tier — what the pending status means for THIS pair:
+#   flat                            : a flat obligation not yet instrumented for recording.
+#   activation_gated_dormant        : a conditional obligation whose runtime is genuinely unwired
+#                                     (its enforcement is dormant; activation_condition unmet).
+#   enforced_pending_instrumentation: the pair IS live-enforced today, but its asserted_obligation_ids
+#                                     RECORDING is deferred to a later slice (see the row's `note`).
+#                                     Parked ≠ unwired for this tier — only the recording is pending.
+#
+# Some `flat` rows carry a `note` from a per-validator enforcement AUDIT explaining why they remain
+# parked (no distinct enforcement site located, or enforcement overlaps an already-recorded
+# obligation) — a surfaced declared≠wired finding, not merely "not yet stamped". The honest rule for
+# RECORDING a pair is that an enforcement-binding test proves a violating input trips a real violation;
+# pairs that fail that bar stay parked rather than being laundered into recorded.
+pending:
+  - validator_id: actionability-matrix-validator
+    obligation_id: reject_matrix_rows_without_baseline_row_or_applicable_validated_expansion_ref
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): the validator strictly requires exactly one resolvable baseline_row_ref per row (the strict check is recorded under baseline-actionability-matrix-validator/reject_matrix_rows_without_baseline_row_ref); the 'OR applicable validated expansion ref' ALTERNATIVE in this current-mode name is NOT implemented — no path admits an expansion-only row. Recording deferred pending that alternative being built or the name reconciled with the strict check."
+  - validator_id: actionability-matrix-validator
+    obligation_id: require_blocker_and_high_rows_to_be_l4_limitation_backed_or_continuation_blocking
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): NO enforcement found in validateActionabilityMatrix — there is no check that blocker/high materiality rows are L4, limitation_backed, or continuation-blocking. A surfaced declared≠wired gap; may be enforced by maturation-continuation-decision-validator / actionable-ontology-validator (owner decision: add here, or re-attribute in the registry)."
+  - validator_id: actionability-matrix-validator
+    obligation_id: support_zero_delta_baseline_only_matrix_when_no_maturation_deltas_exist
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): a 'support' (non-rejection) obligation — satisfied by the ABSENCE of a violation when a matrix mirrors the baseline with no deltas, so there is no bindable violation code to anchor a recording. Stays parked (recording would require an explicit affirmative check)."
+  - validator_id: actionability-matrix-validator
+    obligation_id: validate_maturity_level_against_supporting_refs_blocking_questions_and_limitations
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): enforcement is DISTRIBUTED across regions already owned by other recorded obligations — the member_readiness gate (limitations), the maturity-upgrade citation (supporting_refs; recorded as validate_matrix_rows_derive_from_validated_baseline_and_any_applicable_validated_deltas), and the blocking-question validation (recorded as validate_blocking_question_refs_against_validated_question_frontier). No distinct owner region; recording deferred pending a decision on whether this is a composite obligation or maps to the member_readiness gate specifically."
+  - validator_id: actionable-ontology-validator
+    obligation_id: reject_actionable_ready_when_material_blocker_or_high_question_remains
+    coverage_status: pending
+    tier: flat
+    note: "slice-17 audit (2026-06-23): DELEGATED/NOT_FOUND in this validator. validateActionableOntology directly rejects actionable_ready when final re-question convergence is unproven (recorded separately as reject_actionable_ready_until_final_requestion_convergence_is_proven) and requires the claim to match the continuation decision_state, but it never directly checks that no material blocker/high QUESTION remains open — that closure is enforced by validateMaturationContinuationDecision / the convergence ledger (both required valid here via prior_validation_invalid). Recording deferred pending a direct remaining-blocker/high-question check here, or re-attribution to the continuation-decision validator."
+  - validator_id: actionable-ontology-validator
+    obligation_id: require_static_kinetic_dynamic_surfaces_to_be_closed_or_limitation_excluded
+    coverage_status: pending
+    tier: flat
+    note: "slice-17 audit (2026-06-23): NOT_FOUND for the named per-surface scope. The validator checks per-ROW closure (included projected rows must have matrixRow.member_readiness == closed, ~4996) and full per-row coverage (every matrix row projected exactly once, ~5024), but there is no per-SURFACE check that each of static_surface / kinetic_surface / dynamic_surface is independently closed-or-limitation-excluded (contract: design §Actionability Matrix ~533 'each surface must be independently closed'; continuation-state table ~1116). Surface-level maturity is validated upstream by the actionability-matrix validator, not rechecked here. Recording deferred pending a per-surface closed-or-excluded check (or re-attribution to the matrix validator)."
+  - validator_id: actionable-ontology-validator
+    obligation_id: validate_applicable_proof_authorities_before_projecting_runtime_query_visualization_or_graph_claims
+    coverage_status: pending
+    tier: flat
+    note: "slice-17 audit (2026-06-23): BLANKET-REJECT, not a proof-authority validation. The code rejects ANY downstream_claims value != not_claimed (invalid_enum ~4947-4955), so the active projection can never claim query/visualization/graph-exploration proof. Contract (design ~896-901) says actionable-ontology-validation 'rechecks the same proof authority before projecting' — i.e. it should VALIDATE the applicable proof authority when a downstream surface is claimed, but no proof-authority recheck is implemented (the gate forbids the claim outright rather than validating it). Recording deferred pending real proof-authority rechecking (this is the deferred-capability sibling of the convergence ready-gate)."
+  - validator_id: answer-support-ledger-validator
+    obligation_id: require_contradictions_to_be_recorded_and_bounded
+    coverage_status: pending
+    tier: flat
+    note: "slice-21 audit (2026-06-23): DELEGATED. validateAnswerSupportLedger never reads contradiction_refs. The contract clause (Answer Support Rules: contradictions must prevent answer-claim validation unless the claim records how the contradiction is bounded) is enforced downstream in validateMaturationAnswerClaims (contradictory-cluster -> limitation_refs check), not in this validator. Recording deferred to the answer-claims validator that owns the bounded check."
+  - validator_id: answer-support-ledger-validator
+    obligation_id: require_frontier_triggered_evidence_to_resolve_to_valid_lineage_index_validation
+    coverage_status: pending
+    tier: flat
+    note: "slice-21 audit (2026-06-23): PARTIAL. The valid-lineage-index-validation requirement is an UNSCOPED global precondition (fires when a lineage index validation is present-and-invalid, or when lineage-marked evidence lacks one) rather than scoped to the obligation's frontier_triggered discriminator (membership in the lineage index added_observation_ids). The frontier_triggered scope is read only for the re-entry obligation (recorded); there is no dedicated frontier-triggered-scoped lineage-index-validation enforcement here. Recording deferred."
+  - validator_id: answer-support-ledger-validator
+    obligation_id: validate_external_or_runtime_authority_support_mode
+    coverage_status: pending
+    tier: flat
+    note: "slice-21 audit (2026-06-23): PARTIAL. runtime_proof is presence-only (proof_refs non-empty; the contract's validated query/test/formula/execution/artifact proof that demonstrates the answer is not checked) and authority_response checks response_status == provided but delegates answers-a-closure-frontier-authority-request to maturation-authority-response validation. No distinct name-matching enforcement of the support-mode authority beyond presence/status. Recording deferred."
+  - validator_id: answer-support-ledger-validator
+    obligation_id: validate_support_mode_required_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-21 audit (2026-06-23): AMBIGUOUS/overlapping. The generic required-refs-per-support-mode check is distributed across the per-mode support_mode_missing_authority guards and overlaps the modes owned by validate_user_confirmation_support_mode (user_confirmation) and validate_external_or_runtime_authority_support_mode (runtime_proof/authority_response); only direct_authority's evidence_refs guard is un-owned, but the generic obligation name is not 1:1 with that single mode and cannot be bound non-overlappingly without laundering the others. Recording deferred (would need a dedicated direct-authority-required-refs obligation or a single generic violation code)."
+  - validator_id: answer-support-ledger-validator
+    obligation_id: validate_user_confirmation_support_mode
+    coverage_status: pending
+    tier: flat
+    note: "slice-21 R1 audit (2026-06-23, codex P2): PARTIAL / presence-only. The user_confirmation branch only requires user_confirmation_refs to be non-empty AND a globally-valid purpose-confirmation-validation; it never resolves the cited refs to the confirmation authority, so an arbitrary ref (e.g. ghost-confirmation) passes whenever any valid purpose-confirmation exists. Contrast authority_response, which resolves authority_response_refs to provided responses. Recording deferred pending ref-resolution of user_confirmation_refs against the confirmation authority (WIRE candidate)."
+  - validator_id: baseline-actionability-matrix-validator
+    obligation_id: preserve_frontier_seed_state_before_question_frontier_authoring
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): a defensive (non-mutation) obligation — satisfied by the ABSENCE of post-frontier inputs in baseline mode, with no bindable violation code. Stays parked."
+  - validator_id: baseline-actionability-matrix-validator
+    obligation_id: support_zero_delta_baseline_only_matrix_before_maturation_deltas_exist
+    coverage_status: pending
+    tier: flat
+    note: "slice-3 audit (2026-06-22): a 'support' (non-rejection) obligation with no bindable violation code (the baseline-mode sibling of the current-mode support rule). Stays parked."
+  - validator_id: candidate-disposition-validator
+    obligation_id: require_deferred_or_source_gap_disposition_to_cite_limitation_or_frontier
+    coverage_status: pending
+    tier: flat
+    note: "slice-20 audit (2026-06-23): NOT_FOUND — validateCandidateDisposition never inspects whether a deferred/source_gap disposition cites a limitation or frontier. It checks disposition_id against the registry and requires rationale+evidence, but has no deferred/source_gap-specific limitation/frontier-citation check. Owner: wire the deferred-disposition limitation/frontier requirement."
+  - validator_id: candidate-disposition-validator
+    obligation_id: require_exactly_one_disposition_for_each_salient_candidate
+    coverage_status: pending
+    tier: flat
+    note: "slice-20 audit (2026-06-23): AMBIGUOUS (salience-blind) — the validator enforces exactly-one disposition for EVERY candidate (duplicate_disposition + missing_candidate_disposition), a SUPERSET of the obligation's salient scope, but never reads the `salience` field (ReconstructCandidateInventoryCandidate.salience is a real high/medium/low field, and the contract scopes this to HIGH-salience candidates). Because the validator is salience-blind, it does not validate the obligation's scope discriminator; the named scope has no dedicated enforcer. Recording deferred pending a salience-aware check (or a contract decision to make the requirement salience-agnostic)."
+  - validator_id: candidate-disposition-validator
+    obligation_id: require_high_salience_non_rejected_candidates_to_carry_surface_and_purpose_element_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-20 audit (2026-06-23): NOT_FOUND — the validator never inspects candidate salience, actionability_surface_refs, or purpose_element_refs, so it cannot require high-salience non-rejected candidates to carry surface+purpose-element refs."
+  - validator_id: candidate-disposition-validator
+    obligation_id: validate_candidate_actionability_surface_refs_against_registry_surface_values
+    coverage_status: pending
+    tier: flat
+    note: "slice-20 audit (2026-06-23): NOT_FOUND — the validator never inspects actionability_surface_refs nor validates them against any registry surface catalog. (registryCandidateKindIds/registryCandidateDispositionIds are read, but no surface-value catalog is consulted.)"
+  - validator_id: candidate-disposition-validator
+    obligation_id: validate_candidate_purpose_element_refs_against_selected_purpose_frame
+    coverage_status: pending
+    tier: activation_gated_dormant
+    note: "slice-20 audit (2026-06-23): NOT_FOUND + activation_gated_dormant — registry marks this conditional on activation_condition source_purpose_candidate_runtime_is_implemented, and validateCandidateDisposition takes no purpose-frame input (no source-purpose-candidates-validation arg), so it never validates purpose_element_refs against a selected purpose frame. Stays dormant until that runtime + input wiring exist."
+  - validator_id: claim-projection-validator
+    obligation_id: validate_broader_operated_system_release_health_and_incident_governance_are_bounded
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 audit (2026-06-23): PARTIAL. The name requires BOTH operated_system_release_health AND incident governance to be bounded, but the gate (~913-923, broader_governance_scope_unbounded) only validates governance_scope.operated_system_release_health; the sibling field governance_scope.rollback_quota_incident_governance is never validated anywhere in the validator. Recording deferred pending the gate also bounding rollback_quota_incident_governance."
+  - validator_id: claim-projection-validator
+    obligation_id: validate_material_kind_support_profile_supported_requires_valid_target_material_profile_validation
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 audit (2026-06-23): GATED + INDIRECT. The validity requirement is enforced only as a side-effect: materialKindSupportMemberRows/materialKindSupportProjectionStatus downgrade support to unsupported/blocked when target_material_profile_validation.validation_status != valid (~178/207), and the member-row lineage match (~755-772) compares support_claim. There is no dedicated 'if profile_supported && validation invalid -> violation' check, and the whole block is gated on the optional `args.targetMaterialProfile && args.targetMaterialProfileValidation` (~706) which the validator has no internal guarantee receives. Recording deferred pending a direct, ungated validity check."
+  - validator_id: claim-projection-validator
+    obligation_id: validate_member_capability_rows_preserve_target_ref_profile_snapshot_definition_hash_source_refs_validation_ref_readiness_effect_and_next_action
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 audit (2026-06-23): GATED on optional authority args. The field-by-field preservation check (~755-772, member_capability_lineage_mismatch) is comprehensive and name-matching but only fires inside the `if (args.targetMaterialProfile && args.targetMaterialProfileValidation)` block (~706); both are optional args with no internal guarantee they are supplied (slice-18/22 caller-supplied-optional principle). Recording deferred pending an internal guarantee the target-profile authority is always present (or a conditional-stamp + self-test design)."
+  - validator_id: claim-projection-validator
+    obligation_id: require_blocked_projection_to_have_limitation_or_recovery_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 R1 audit (2026-06-23, codex P2): subset of named scope. The unconditional guard (~924-934, blocked_projection_missing_recovery_ref) keys only on claim_level=='blocked', but material_kind_support rows express blocked via machine_status=='blocked' while keeping claim_level=='not_applicable'. Their blocked-recovery check (~893-912) is gated on the optional requiredValidationStatuses arg, so when it is absent a blocked material-support row can drop its limitation refs and pass. Recording deferred pending the unconditional check also covering material_kind_support machine_status=='blocked'."
+  - validator_id: claim-projection-validator
+    obligation_id: require_each_projection_row_to_cite_runtime_validation_authorities
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 R1 audit (2026-06-23, codex P2): presence-only proxy. The unconditional check (~861-867, required_validation_ref_missing) only requires required_validation_refs to be non-empty; arbitrary strings (e.g. ghost-validation.yaml) pass. Whether the refs are the ACTUAL runtime validation authorities is checked only when the optional expectedRequiredValidationRefs is supplied (~868-892), which the validator does not internally derive. Recording deferred pending a mandatory/derived authority-set match (cf. slice-21 user_confirmation presence-only park)."
+  - validator_id: claim-projection-validator
+    obligation_id: validate_all_public_and_downstream_projection_surfaces_are_present
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 R1 audit (2026-06-23, codex P2): presence-only, not one-per-surface. The check (~949-956, missing_required_surface) verifies each PROJECTION_SURFACE appears at least once, but the contract (design ~1172) requires every status/result/API/MCP/handoff/material-kind-support claim to consume ONE pre-publication projection row. A duplicate surface (two rows, unique ids) passes the presence check. The one-row-per-surface guarantee is enforced only via the gated expectedClaimProjection derived-match. Recording deferred pending an unconditional one-row-per-surface check."
+  - validator_id: claim-projection-validator
+    obligation_id: validate_claim_level_decision_state_and_actionability_claim_alignment
+    coverage_status: pending
+    tier: flat
+    note: "slice-23 R1 audit (2026-06-23, codex P2): one-directional. The unconditional checks (~829-860) enforce decision_state=>actionability_claim (actionable_ready=>ready, actionable_limited=>limited) and claim_level actionable_ready=>actionable_ready decision, but NOT the reverse: a row with decision_state=='continue', claim_level=='maturation_in_progress', actionability_claim=='ready' is accepted. Full alignment is enforced only via the gated expectedClaimProjection derived-match. Recording deferred pending bidirectional alignment (actionability_claim=='ready' requires actionable_ready decision_state)."
+  - validator_id: competency-question-assessment-validator
+    obligation_id: validate_answer_status_against_active_answerability_contract
+    coverage_status: pending
+    tier: flat
+    note: "slice-12 audit (2026-06-22): AMBIGUOUS — structure-only, content judgment is blind. The validator checks answer_status enum validity (invalid_enum) + per-status STRUCTURAL requirements (answerable→evidence_refs + required_seed_refs; unsupported/deferred→missing_source_or_confirmation), but never verifies the answer_status reflects actual answerability from evidence CONTENT — that judgment is the content-blind host-LLM assessment step (see contract-runtime-gap-ledger silent-defect #1), not validated here. The per-status structural rules are real but the name (\"active answerability contract\") implies the content/answerability semantics the validator does not check. Recording deferred pending content-aware answerability validation or name reconciliation."
+  - validator_id: competency-question-assessment-validator
+    obligation_id: validate_answerability_trace_refs_close_against_seed_evidence_limitations_and_proofs
+    coverage_status: pending
+    tier: flat
+    note: "slice-12 audit (2026-06-22): NOT_FOUND for the named arms. The assessment artifact has no answerability_trace_refs, limitation_refs, or proof_refs fields. The validator closes evidence_refs (unknown_observation_ref) and required_seed_refs (recorded separately as validate_required_seed_refs_close_against_question_seed_refs) and linked_claim_ids against the question, but NEVER against limitations or proofs (proofs are separate query/visualization/graph-exploration-proofs artifacts). Of the four named arms (seed, evidence, limitations, proofs) only seed+evidence are closed, and there is no trace_refs field. Recording deferred pending limitation/proof closure (or name reconciliation; the seed arm is already covered)."
+  - validator_id: competency-questions-validator
+    obligation_id: emit_required_evidence_scope_runtime_projection
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit (2026-06-23): PROJECTION_ONLY — the validator derives derivedRequiredEvidenceScope(question) per question and emits it to required_evidence_scope_projection[] in the return (post-seed-validation.ts ~1451), with NO violation. A value derived/exposed in the output with no breaching binding is unrecordable (slice-24 projection-only kind); the consumer that acts on the projection would own any enforcement."
+  - validator_id: competency-questions-validator
+    obligation_id: enforce_p3_diagnostic_or_claim_based_admission_disposition_rule
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit (2026-06-23): NOT_FOUND — the validator has no code enforcing a P3/diagnostic admission-disposition rule (no priority or diagnostic-disposition field is read against the trace-ref membership). The admitted-disposition enforcement that IS wired is the exactly-one-per-admitted-id rule (recorded as enforce_admitted_domain_competency_disposition_rule). No bindable violation for the P3-diagnostic arm; recording deferred pending implementation."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_classification_level_axis_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit (2026-06-23): NOT_FOUND — the per-question loop never reads a classification_level_axis_facets[] field; no such field exists on the competency-question schema (classification_level_axis_refs[] exists only on ontology-seed.classification_mapping, not on questions). declared≠wired; recording deferred pending the field being added and read."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_hidden_label_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit (2026-06-23): NOT_FOUND — the per-question loop never reads a hidden_label_facets[] field; the contract (§15) folds hidden-label/search-label policy into terminology coverage (terminology_facets[], recorded as validate_terminology_facet_coverage). No separate hidden_label field exists or is read; recording deferred pending name reconciliation or a dedicated field."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_query_visualization_graph_contract_refs_when_applicable
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit (2026-06-23) + codex #142: PARTIAL/presence-only — the validator resolves PRESENT query_access/visualization/graph_exploration_contract_refs against the registry (unknown_id), but never enforces the '_when_applicable' arm: when a seed makes a query/visualization/graph contract applicable, the matching refs must be present (or limitation-backed), and an empty applicable array is silently accepted. The ref-resolution half is wired but the applicability-presence half is not, so the named scope is not fully enforced. Recording deferred pending the presence check (a behavior change beyond instrumentation)."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_reasoning_formalism_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — the validator only validateRefArray-resolves the PRESENT reasoning_or_formalism_facets refs against the registry (unknown_id); it never enforces the contract '_coverage' arm (when ontology_handoff_axis_refs includes reasoning_or_formalism_profile, the question must cover or limitation-back the reasoning facets). The conditional facet-presence requirement is unwired, so the named coverage scope is not enforced. Recording deferred pending the per-axis facet-presence check."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_entity_identity_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT entity_identity_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the entity_identity axis is applicable. Same presence-only gap as the sibling *_facet_coverage obligations. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_instance_assertion_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT instance_assertion_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the instance_assertion axis is applicable. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_terminology_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT terminology_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the terminology axis is applicable. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_relation_type_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT relation_type_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the relation_typing axis is applicable. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_classification_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT classification_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the classification axis is applicable. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_constraint_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL/presence-only — only resolves PRESENT constraint_facets refs (unknown_id); never enforces the conditional facet-presence requirement when the constraints axis is applicable. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_modeling_concern_facet_coverage
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL — the per-question modeling_concern_facets check is presence-only (resolves present refs, unknown_id), and the post-loop required-concern coverage check (requiredModelingConcerns must be covered) populates coveredModelingConcerns from ANY question regardless of lifecycle_status, so an inactive/unsupported question counts as coverage (same inactive-coverage gap codex flagged for the required-axis obligation). Neither arm fully enforces the named coverage scope. Recording deferred."
+  - validator_id: competency-questions-validator
+    obligation_id: validate_required_ontology_handoff_axis_coverage_or_limitation_backed_disposition
+    coverage_status: pending
+    tier: flat
+    note: "slice-27 audit + codex #142 R2: PARTIAL — coveredOntologyHandoffAxes is populated from EVERY question's ontology_handoff_axis_refs before any lifecycle/disposition filtering, so a required axis mentioned only by an inactive (lifecycle_status=unsupported_candidate) or non-limitation-backed question still satisfies the post-loop coverage check. The obligation requires representation by an ACTIVE question OR an unsupported/deferred limitation-backed disposition; that active/limitation-backed filter is unwired. Recording deferred."
+  - validator_id: failure-classification-validator
+    obligation_id: classify_missing_or_failed_active_validation_artifact_as_required_gate_failure
+    coverage_status: pending
+    tier: flat
+    note: "slice-13 audit (2026-06-22, codex PR #123): UNWIRED ON THIS VALIDATOR (not a re-attribution). The obligation is registered on failure-classification-validator and design.md:3275-3280 says failure-classification VALIDATION must consume the failed-gate validation artifacts (and per the failure_classification_gate required_when predicate, runtime-halt / unresolved-work evidence such as reconstruct-metrics.yaml, not only missing/failed validations). But validateFailureClassification (post-seed-validation.ts) takes only failure_classification + competency-question-assessment + seed-confirmation-validation and validates authored rows structurally (duplicate_id / failure_kind invalid_enum / question_id+claim_id unknown_id / rationale_missing — none a registered obligation); it never reads the gate validation artifacts to detect missing/failed/halted state and classify them as required-gate failures. Backlog: wire this consumption INTO this validator; the AUTHORITATIVE input set is the registry record's input_authority_refs + dynamic_input_authority_rule + the gate's required_when predicate (do not rely on this note's prose as the spec — read the registry). Recording deferred."
+  - validator_id: failure-classification-validator
+    obligation_id: classify_missing_or_failed_promoted_maturation_validation_artifact_as_required_gate_failure
+    coverage_status: pending
+    tier: flat
+    note: "slice-13 audit (2026-06-22, codex PR #123): UNWIRED ON THIS VALIDATOR. The registry binds the promoted/requested validations to this validator via conditional_input_authority_refs + the dynamic_input_authority_rule, but validateFailureClassification receives none of them and performs no promoted-gate missing/failed classification. Backlog: wire INTO this validator over the registry-declared promoted-authority set (the registry's conditional_input_authority_refs/dynamic rule is the authoritative list — do not enumerate from this note). Recording deferred."
+  - validator_id: failure-classification-validator
+    obligation_id: derive_planned_gate_failure_inputs_from_registry_gate_catalogs_not_hand_maintained_lists
+    coverage_status: pending
+    tier: flat
+    note: "slice-13 audit (2026-06-22, codex PR #123): UNWIRED ON THIS VALIDATOR. The registry's dynamic_input_authority_rule explicitly assigns THIS validator to 'consume every validation_artifact_ref from validation_gate_catalog and planned_validation_gate_catalog whose required_when is true … hand-maintained conditional_input_authority_refs are non-authoritative cache hints only' (design.md:2336-2340). validateFailureClassification never receives the registry/catalogs and does no required_when-driven derivation. Backlog: wire the catalog-driven derivation INTO this validator. Recording deferred."
+  - validator_id: failure-classification-validator
+    obligation_id: validate_failure_trigger_against_active_runtime_validation_authorities
+    coverage_status: pending
+    tier: flat
+    note: "slice-13 audit (2026-06-22, codex PR #123): AMBIGUOUS — name broader than code. The validator resolves failure.question_id against the competency-question-assessment and failure.claim_id against the seed-confirmation-validation (unknown_id) — the failure's SUBJECT refs, not its TRIGGER (the missing/failed gate validation artifact the contract names). The registry declares the active runtime VALIDATION-artifact authorities for this validator (a subset of input_authority_refs — NOT the authored failure-classification.yaml, run-manifest, or registry metadata refs); the function reads only two of them. Backlog: wire a trigger check over the registry-declared active validation-artifact authorities (read the registry for the exact refs). Recording deferred."
+  - validator_id: failure-classification-validator
+    obligation_id: validate_failure_trigger_against_promoted_planned_validation_authorities
+    coverage_status: pending
+    tier: flat
+    note: "slice-13 audit (2026-06-22, codex PR #123): UNWIRED ON THIS VALIDATOR over the promoted/requested authority set. validateFailureClassification receives none of the promoted-planned validation authorities. Backlog: wire the trigger check over the AUTHORITATIVE promoted-authority set = the registry record's conditional_input_authority_refs whose consumed_for lists this obligation, resolved via the dynamic_input_authority_rule (read the registry for the exact refs — this note deliberately does not enumerate them, since the registry is the SSOT and a prose list drifts). Recording deferred."
+  - validator_id: handoff-decision-validator
+    obligation_id: consume_all_requested_or_promoted_planned_gate_statuses_emitted_by_runtime
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): NOT_FOUND — depends on planned_validation_gate_catalog, which is declared in the registry YAML but NEVER loaded into the runtime ReconstructContractRegistry (the loader does not parse it; grep: zero references to planned_validation_gate_catalog in src/core-runtime/reconstruct). projectGateStatusesOnce iterates contractRegistry.validation_gate_catalog (active gates) only, so requested/promoted planned gate statuses are never consumed. This is a genuine declared-but-unwired surfacing (owner: load planned gates into the registry, or re-scope the obligation)."
+  - validator_id: handoff-decision-validator
+    obligation_id: derive_terminal_planned_gate_inputs_from_registry_gate_catalogs_not_hand_maintained_lists
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): NOT_FOUND — same root cause as consume-planned: planned_validation_gate_catalog is never loaded, so terminal planned gate inputs are not derived from any registry catalog. The active gate projection derives from validation_gate_catalog, but the planned-gate derivation this obligation names does not exist in the runtime."
+  - validator_id: handoff-decision-validator
+    obligation_id: do_not_require_failure_or_revision_artifacts_for_clean_stop_decisions
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): defensive NEGATIVE obligation — the validator does not require failure-classification/revision-proposal artifacts for a clean stop (those gates are conditional via registry required_when predicates, skipped when not_applicable). There is no positive bindable violation for 'do not require' (a clean stop simply clears); the property is delegated to the registry predicate catalog, not a distinct enforcer in validateHandoffDecision. Recording deferred (no non-vacuous breaching binding)."
+  - validator_id: handoff-decision-validator
+    obligation_id: project_missing_required_promoted_or_requested_planned_gate_validation_artifact_as_blocked
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): NOT_FOUND — depends on planned_validation_gate_catalog (never loaded). Missing planned gate validation artifacts are never consulted or projected as blocked; only missing ACTIVE artifacts are (recorded as project_missing_active_validation_artifact_as_blocked)."
+  - validator_id: handoff-decision-validator
+    obligation_id: validate_accepted_maturation_execution_minimum_output_gates_are_terminal_projected
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): AMBIGUOUS — no distinct 'accepted maturation execution minimum output gates' selection exists. projectGateStatusesOnce terminal-projects ALL applicable active gates from validation_gate_catalog, but there is no maturation-minimum filter or maturation-context gate selection naming this obligation, and the planned/maturation gates it implies are unloaded. No name-matching enforcer."
+  - validator_id: handoff-decision-validator
+    obligation_id: validate_failure_classification_covers_failed_or_missing_requested_planned_gates
+    coverage_status: pending
+    tier: flat
+    note: "slice-19 audit (2026-06-23): NOT_FOUND — depends on planned_validation_gate_catalog (never loaded). The validator consumes failureClassificationValidation only as metadata (material_failure_count) and for predicate evaluation; it never validates that failure classification COVERS failed/missing planned gates."
+  - validator_id: material-admission-ledger-validator
+    obligation_id: validate_phase_scoped_input_kind_disposition_and_snapshot_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-15 audit (2026-06-23): AMBIGUOUS — multi-facet name, one facet unenforced. The name binds THREE facets: phase-scoped input_kind, phase-scoped disposition, and snapshot_refs. Two are enforced — phase→input_kind (invalid_phase_input_kind: pre_seed_purpose_element→purpose_adequacy_element, pre_seed_material_value→material_value, post_cq_domain_competency→domain_competency_question) and input_kind→snapshot_ref (missing_snapshot_ref per input_kind). But the DISPOSITION facet is NOT enforced: the code only validates disposition against the global enum (invalid_enum), never restricting disposition VALUES by input_kind/admission_phase. The contract (ontology-seeding-and-maturation-design.md §Material value and domain competency admission, the admit-when table) restricts dispositions per input_kind (material_value ⊆ {admitted_material, trace_audit_only, out_of_scope, deferred_authority, rejected_ambiguous}; domain_competency_question ⊆ {required_blocking, supporting_material, diagnostic_only, deferred_product_decision, out_of_scope}), so e.g. a material_value row carrying required_blocking passes uncaught. Recording deferred pending a per-input_kind disposition allow-set check (or the name reconciled to drop the disposition facet)."
+  - validator_id: maturation-answer-claims-validator
+    obligation_id: require_partial_answers_to_have_limitation_or_frontier_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-4 audit (2026-06-22): validateMaturationAnswerClaims enforces 'partially_answered → limitation_refs' (missing_required_ref), but the 'OR frontier' alternative is unrepresentable — ReconstructMaturationAnswerClaim has no frontier_refs field. Name/artifact mismatch; recording deferred pending the frontier alternative being added or the name reconciled with the limitation-only check."
+  - validator_id: maturation-answer-claims-validator
+    obligation_id: validate_answer_claim_surface_dimension_and_purpose_element_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-4 audit (2026-06-22): all three ref arrays are presence-validated (missing_required_coverage) and target_surface_refs are resolved against the question (conflicting_state), but target_dimension_refs and purpose_element_refs are NOT resolved against the question — only presence-checked. Recording deferred pending dimension/purpose resolution checks or an owner decision (partial enforcement)."
+  - validator_id: maturation-answer-claims-validator
+    obligation_id: validate_support_mode_against_valid_evidence_cluster_or_authority
+    coverage_status: pending
+    tier: flat
+    note: "slice-4 audit (2026-06-22): support_mode is validated against evidence clusters (invalid_enum, conflicting_state on cluster-mode mismatch, support_mode_missing_authority when no cluster is cited), but the 'OR authority' alternative branch is not implemented — there is no validation path against an authority artifact. Recording deferred pending the authority branch or name reconciliation."
+  - validator_id: maturation-authority-response-validator
+    obligation_id: preserve_unavailable_or_rejected_authority_as_blocked_or_limitation_state
+    coverage_status: pending
+    tier: flat
+    note: "slice-14 audit (2026-06-22): NOT_FOUND in validateMaturationAuthorityResponse (maturation-validation.ts). unavailable/deferred responses are only COUNTED into unavailable_response_count (return field); no violation fires when an unavailable/rejected/deferred/contradicted response fails to be preserved as a blocked/limitation state (no limitation_refs requirement). Preserving the gap as blocked/limitation is a downstream continuation/closure-decision consumer's job. Recording deferred pending a validator-side preservation check."
+  - validator_id: maturation-authority-response-validator
+    obligation_id: reject_authority_response_that_claims_source_support_without_source_observation_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-14 audit (2026-06-22, codex PR #124): a substring PROXY, not resolution. The only check is supporting_refs.some(ref => ref.includes('source-observations')) (support_mode_missing_authority) — it never loads source-observations.yaml or resolves observation ids, and a misleading ref such as 'not-source-observations' satisfies the substring while citing no real source observation. A breaching input (claims source support, no real source-observation ref) escapes, so this is not contract-complete. Recording deferred pending resolution against source-observations.yaml."
+  - validator_id: maturation-authority-response-validator
+    obligation_id: validate_authority_identity_and_snapshot_refs_are_recorded
+    coverage_status: pending
+    tier: flat
+    note: "slice-14 audit (2026-06-22): AMBIGUOUS — only the identity arm is enforced. The validator requires authority_identity.authority_id/label/role to be non-empty (missing_required_ref), but never checks authority_snapshot_ref or authority_version_or_timestamp are recorded. Name says 'identity AND snapshot refs'; the snapshot arm is unenforced. Recording deferred pending a snapshot-ref recorded check."
+  - validator_id: maturation-authority-response-validator
+    obligation_id: validate_authority_response_refs_against_maturation_closure_frontier_authority_requests
+    coverage_status: pending
+    tier: flat
+    note: "slice-14 audit (2026-06-22, codex PR #124): the validator does resolve authority_request_ref against maturationClosureFrontier.authority_requests (unknown_id) + kind match (conflicting_state), but recording it is unsafe on two counts. (1) The validator record under-declares its inputs — it lists maturation-authority-response.yaml + maturation-closure-frontier-VALIDATION.yaml + registry, but NOT maturation-closure-frontier.yaml, the artifact whose authority_requests the check needs; a registry-driven consumer would not provide it (declared≠wired). (2) No cross-artifact binding: the validator resolves against whichever frontier the caller passes and never checks response.closure_frontier_ref or frontier-validation.maturation_closure_frontier_ref name that same frontier, so a reused authority-response from another frontier validates against a mismatched frontier with matching ids (same class as the #100 matrix cross-artifact-binding follow-up). Recording deferred pending the frontier-artifact input declaration + response↔frontier↔frontier-validation binding."
+  - validator_id: maturation-authority-response-validator
+    obligation_id: validate_authority_response_status_and_scope
+    coverage_status: pending
+    tier: flat
+    note: "slice-14 audit (2026-06-22): AMBIGUOUS — name broader than code. The only status-specific enforcement is response_status=='provided' requiring supporting_refs or response_source_ref (missing_required_ref); the other statuses (unavailable/rejected/deferred/contradicted) carry no status validation, and there is no explicit 'scope' check (the authority_kind match belongs to validate_authority_response_refs_against_maturation_closure_frontier_authority_requests). Recording deferred pending full status/scope validation."
+  - validator_id: maturation-baseline-validator
+    obligation_id: preserve_immutable_source_seed_and_validation_refs_used_to_compute_baseline
+    coverage_status: pending
+    tier: flat
+    note: "slice-2 audit (2026-06-22): validateMaturationBaseline copies the source seed/validation refs into its validation artifact but enforces no immutability/tamper check on them — no distinct enforcement site located. Recording deferred pending an added enforcement check or an owner decision (the obligation may be enforced by a downstream validator and mis-attributed here)."
+  - validator_id: maturation-baseline-validator
+    obligation_id: validate_baseline_rows_cover_static_kinetic_dynamic_surfaces_and_maturity_dimensions
+    coverage_status: pending
+    tier: flat
+    note: "slice-2 audit (2026-06-22): no enforcement site DISTINCT from the already-recorded required-element tuple coverage (validate_baseline_rows_cover_selected_purpose_frame_required_elements) was located; the per-row surface/dimension checks validate present-row validity, not full static/kinetic/dynamic surface coverage. Recording deferred pending a distinct site or an owner decision to fold this into the recorded coverage obligation."
+  - validator_id: maturation-baseline-validator
+    obligation_id: validate_baseline_rows_trace_to_seed_refs_limitations_or_validated_assessment_gaps
+    coverage_status: pending
+    tier: flat
+    note: "slice-2 audit (2026-06-22): validateMaturationBaseline validates purpose_element_ref resolution but enforces no per-row trace-to-seed_refs/limitations/validated-assessment-gaps check — no distinct enforcement site located. Recording deferred pending an added enforcement check or an owner decision."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: reject_duplicate_or_already_observed_source_requests
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): PARTIAL. The validator de-dupes source_request_id and rejects already-observed refs, but two requests with DISTINCT ids targeting the same unobserved requested_source_ref both enter accepted_source_request_ids. In a real run that passes validation and then the re-entry observer throws on the second accepted id (run.ts ~9870-9884). Contract (design ~1372-1373) requires rejecting duplicate source requests (same target). Recording deferred pending same-target-ref duplicate rejection."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: reject_unsupported_material_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): PARTIAL. The validator treats ANY inventory row as supported and only rejects a missing row; rows with exists:false, scan_status:skipped, or a non-concrete kind still validate and are accepted. The runtime then builds a detection and buildReconstructSourceObservation returns null for non-existent/non-concrete refs, throwing (materialize-preparation.ts ~434-436; run.ts ~9908-9911). Recording deferred pending rejection of skipped/non-existent/non-concrete inventory entries."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: validate_closure_frontier_question_refs_against_material_unanswered_questions
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): PARTIAL + declared≠wired. (a) An empty source-request question_refs array fires no violation, so a request resolves to zero material questions and still validates. (b) The check reads questions/materialQuestions from the maturation-question-frontier ARTIFACT, but the registry input_authority_refs (registry ~2613-2619) declares only maturation-question-frontier-validation.yaml, not the artifact — a registry-driven consumer cannot supply it (slice-14 pattern). Recording deferred pending non-empty-refs enforcement + the frontier artifact declared as an input."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: validate_closure_frontier_source_requests_preserve_member_and_cross_material_lineage
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): dead discriminator. The mixed-lineage check fires only when an individual source request has target_material_kind === 'mixed', but real mixed targets are an aggregate detection while inventory units / source requests carry concrete per-ref kinds (materialize-preparation.ts ~46-56, ~729-732) that exclude 'mixed'. A mixed target requesting a concrete member source (e.g. code) with empty lineage validates anyway. Contract (design ~1374-1376) requires lineage for mixed-material source requests. Recording deferred pending enforcement keyed on the real (aggregate) mixed-target signal or the target profile."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: validate_authority_request_question_refs_against_material_unanswered_questions
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): PARTIAL + declared≠wired. Same two issues as the source-request question_refs obligation: an empty authority-request question_refs array fires no violation, and the materiality check reads the maturation-question-frontier ARTIFACT which the registry does not declare as an input authority (registry ~2613-2619). Recording deferred pending non-empty-refs enforcement + the frontier artifact declared as an input."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: validate_authority_request_blocking_semantics_against_question_materiality
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): declared≠wired. The blocker-must-block check reads question.materiality from the maturation-question-frontier ARTIFACT, but the registry input_authority_refs (registry ~2613-2619) declares only maturation-question-frontier-validation.yaml, not the artifact — a registry-driven consumer cannot reproduce the check (slice-14 pattern). Recording deferred pending the frontier artifact declared as an input authority."
+  - validator_id: maturation-closure-frontier-validator
+    obligation_id: reject_duplicate_authority_requests_for_same_question_authority_kind_and_scope
+    coverage_status: pending
+    tier: flat
+    note: "slice-25 R1 audit (2026-06-23, codex P2): scope-too-coarse. The dedupe keys on the ENTIRE sorted question_refs array, so a request covering [q1,q2] and another covering [q1] with the same authority kind and scope both validate even though they duplicate the same (question, kind, scope) for q1. Contract (design ~1379-1380) is per-question. Recording deferred pending per-question dedupe."
+  - validator_id: maturation-continuation-decision-validator
+    obligation_id: validate_continuation_state_against_validated_actionability_matrix_and_frontier_state
+    coverage_status: pending
+    tier: flat
+    note: "slice-22 audit (2026-06-23): AMBIGUOUS. The name is broad ('continuation_state against validated matrix AND frontier_state') and its intent is split across three non-overlapping enforcement regions with no single block that matches it 1:1: the blanket prior-validation loop (requires all 8 upstream validations valid, of which matrix/question-frontier/closure-frontier are three), the claim_scope partition recompute (decision included/excluded refs must equal the matrix closed/non-closed rows — overlaps the actionable_limited row-scope obligation), and the material-open-rows gate (overlaps reject_actionable_ready_when_material_blocker_or_high_row_remains_unclosed). Recording deferred (would need splitting into narrower obligations e.g. require_prior_validation_status_valid / validate_claim_scope_partition_against_matrix, or absorbing into the existing recorded gates)."
+  - validator_id: maturation-continuation-decision-validator
+    obligation_id: validate_continue_ask_user_or_blocked_state_against_available_next_authority
+    coverage_status: pending
+    tier: flat
+    note: "slice-22 audit (2026-06-23): PARTIAL. The validator enforces continue (requires >=1 next_frontier_ref) and ask_user (requires >=1 authority_request_ref), but the named `blocked` state has NO enforcement against next-authority availability. Contract (design ~1539: deferred_external_authority projects ask_user/blocked/limitation depending on availability) expects blocked to be validated against unavailability, but no such check exists. Recording deferred pending a `blocked`-state enforcer."
+  - validator_id: maturation-continuation-decision-validator
+    obligation_id: bind_revision_proposal_validation_to_consumed_revision_proposal
+    coverage_status: pending
+    tier: flat
+    note: "slice-22 R1 audit (2026-06-23, codex P2): caller-supplied-optional-arg gate. The binding check (revisionProposalValidation.revision_proposal_ref must match the consumed proposal) only runs when the caller supplies the optional `revisionProposalRef`; the validator has no internal guarantee it is provided, so for direct/pure calls the obligation is advertised but the check is inert. Same disposition as slice-18 expectedRegistrySha256 (run-control slice-10 principle). Recording deferred pending an internal guarantee that the consumed proposal ref is always present."
+  - validator_id: maturation-continuation-decision-validator
+    obligation_id: reject_actionable_ready_when_material_blocker_or_high_row_remains_unclosed
+    coverage_status: pending
+    tier: flat
+    note: "slice-22 R1 audit (2026-06-23, codex P2): subset of named scope. The gate builds materialOpenRows from blocker/high rows with member_readiness=='frontier_required' only, but the obligation names 'unclosed' which also covers 'limitation_backed' (a distinct non-closed member_readiness value). A hand-edited actionable_ready decision with a limitation_backed blocker/high row therefore bypasses the gate. Recording deferred pending the gate covering all non-closed material rows (frontier_required + limitation_backed)."
+  - validator_id: maturation-continuation-decision-validator
+    obligation_id: validate_limitation_refs_and_row_scope_for_actionable_limited_state
+    coverage_status: pending
+    tier: flat
+    note: "slice-22 R2 audit (2026-06-23, codex P2): name-broader-than-code. The validator firmly enforces the row_scope facet (actionable_limited requires >=1 included_row_ref) but the limitation-ref facet is only a weak presence check (requires excluded_row_refs OR limitation_refs to be non-empty). An actionable_limited decision that excludes a limitation_backed matrix row but DROPS that row's limitation_refs still passes (satisfied by the excluded row alone). Only candidate-limitation and revision-blocker preservation are enforced (separate obligations); per-row limitation_backed limitation_refs are not preserved. Recording deferred pending row-level limitation-ref preservation."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: preserve_remaining_frontier_refs_without_ready_projection
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23): AMBIGUOUS — composite name, only the resolution sub-check is enforced. The code validates remaining_frontier_refs resolve to frontier questions (unknown_id, maturation-validation.ts ~3874) = the FIRST clause of contract bullet (h) (design §convergence-ledger must-prove). But the 'without ready projection' framing references the ready-projection terminal gate that contract (h) explicitly DEFERS ('becomes a stronger terminal projection gate WHEN actionable ontology projection is implemented'); this validator has no ready-projection logic and no enforcer ties remaining-frontier preservation to a ready gate. A breaching input binds only the resolution check, not the composite. Recording deferred pending the ready-projection gate, or re-attribution as a plain resolve-remaining-frontier-refs obligation."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: reject_actionable_ready_until_final_requestion_convergence_is_proven
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23): NOT_FOUND in this validator — DEFERRED gate. The convergence validator only REPORTS final_requestion_pass_status (it surfaces it on the output) and checks the pass-status/new-material-ref consistency; it never gates an actionable_ready projection. Contract bullet (h) states the ready gate 'becomes a stronger terminal projection gate WHEN actionable ontology projection is implemented' — i.e. rejecting actionable_ready until convergence is proven is owned by the actionable-ontology / continuation-decision projection, not here. Recording deferred pending that projection."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: reject_trace_audit_or_authority_request_rows_as_positive_semantic_support
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23): AMBIGUOUS — no name-matching positive-support enforcer. The closure-disposition enum has no 'authority_request' value (closest: deferred_external_authority), and the validator never rejects trace_audit_only or deferred rows from being counted as positive semantic support. The nearest check (conflicting_state ~3839) forbids trace_audit_only rows from citing expansions or closing blocker/high questions = contract bullet (d), a DIFFERENT rule. positiveSupportRefs (~3799) is computed and consumed only by the answered_no_semantic_change branch, not by a trace/authority exclusion gate. Recording deferred pending an explicit positive-support exclusion check (or name reconciliation against the trace_audit_only rule that IS enforced)."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: require_every_blocker_or_high_question_to_have_answer_expansion_blocked_or_frontier_closure
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23, codex PR #127): PARK — the recorded enforcer would be a weak proxy. Contract bullet (a) requires every blocker/high question to be closed, CARRIED FORWARD, or blocked WITH REFS. The code (missing_required_coverage, maturation-validation.ts ~3763) only checks the question id appears in some closure_rows[].question_refs: it (1) rejects the contract-permitted carried-forward-via-remaining_frontier_refs path, and (2) accepts a dummy out_of_scope/low closure row with no refs as 'closure' — it never validates the permitted dispositions or refs. Recording would let G10 treat the full closed/carried-forward/blocked-with-refs contract as covered. Recording deferred pending a check that validates the permitted closure dispositions + refs (and accepts carry-forward)."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: validate_each_source_observation_delta_row_has_convergence_closure_disposition
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23, codex PR #127): PARK — two gaps. (1) The closure-coverage enforcer (missing_required_coverage ~3710) is INSIDE the per-round loop; a source-observation-delta artifact with delta rows but an empty rounds[] array enforces nothing, so a stamp gated only on sourceObservationDelta presence would launder the zero-round shape. (2) Contract bullet (g) requires each consumed delta row to appear in exactly one source_observation_closure_rows[] row WITH A DISPOSITION, but the source-closure branch never reads sourceClosure.closure_disposition — a missing/out-of-enum disposition still passes. Recording deferred pending (a) gating on a reached source-delta round and (b) validating the closure_disposition field."
+  - validator_id: maturation-convergence-ledger-validator
+    obligation_id: validate_prior_maturation_validations_are_valid
+    coverage_status: pending
+    tier: flat
+    note: "slice-16 audit (2026-06-23, codex PR #127): PARK — incomplete prior-validation scope. The code checks the validation_status of the question-frontier, actionability-matrix, answer-support, answer-claims, and ontology-expansion validations (prior_validation_invalid ~3500-3538). But the registry declares rounds/<round-id>/source-observation-delta-validation.yaml as an input authority for delta-backed rounds, and this validator never receives or checks that validation artifact's status — an invalid source-observation-delta validation can be referenced while the obligation claims prior validations were proven. (Separately, actionability-matrix-validation is checked by code but NOT declared in the registry input_authority_refs — an inverse declared≠wired the owner may reconcile.) Recording deferred pending the source-delta validation status being added to the gate (or recording gated to non-delta runs)."
+  - validator_id: maturation-question-frontier-validator
+    obligation_id: validate_authority_need_rows_are_first_class_and_scope_to_questions
+    coverage_status: pending
+    tier: flat
+    note: "slice-6 audit (2026-06-22): validateMaturationQuestionFrontier reads question.authority_need only as part of the blocker/high closure check (recorded as require_blocker_and_high_questions_...); there is NO independent enforcement that authority_need rows are first-class or scoped to questions (authority_scope / expected_response_kind / blocking semantics are not validated here). A surfaced declared≠wired gap (may be enforced by maturation-closure-frontier-validator). Recording deferred."
+  - validator_id: maturation-question-frontier-validator
+    obligation_id: validate_competency_question_assessment_and_domain_trace_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-6 audit (2026-06-22): NO enforcement found — competency_question_refs / competency_assessment_refs / domain_competency_trace_refs are not accessed or resolved in validateMaturationQuestionFrontier. A surfaced declared≠wired gap. Recording deferred pending an added check or registry re-attribution."
+  - validator_id: maturation-question-frontier-validator
+    obligation_id: validate_question_materiality_surface_dimension_and_answer_status_enums
+    coverage_status: pending
+    tier: flat
+    note: "slice-6 audit (2026-06-22): materiality (invalid_enum) and surface/dimension non-emptiness (missing_required_coverage) ARE enforced, but the named answer_status enum clause is NOT — question.current_answer_status is never validated as an enum here. Partial; recording deferred pending the answer_status check or name reconciliation."
+  - validator_id: maturation-question-frontier-validator
+    obligation_id: validate_question_refs_against_baseline_seed_and_limitation_authorities
+    coverage_status: pending
+    tier: flat
+    note: "slice-6 audit (2026-06-22): baseline_row_refs ARE resolved against the maturation baseline (unknown_id) and purpose_element preservation is checked, but the named SEED and LIMITATION authority clauses are NOT enforced — seed_ref_refs are not accessed and limitation_refs are only checked for non-emptiness (as part of the closure obligation), never resolved against an authority. Partial; recording deferred."
+  - validator_id: ontology-expansion-validator
+    obligation_id: prevent_in_place_seed_authority_rewrite
+    coverage_status: pending
+    tier: flat
+    note: "slice-9 audit (2026-06-22, codex PR #119 finding): the contract clause is unscoped ('no expansion rewrites seed authority in place'), but the check only rejects refs whose path.basename(ref) === 'ontology-seed.yaml' (seed_authority_rewrite_attempt). target_seed_or_ontology_refs are LLM-authored free strings with no normalization (run.ts stringArray), so an anchored ref such as 'ontology-seed.yaml#semantic_layer/object-new' or 'ontology-seed.yaml:semantic_layer' has a different basename and bypasses the check while still naming the seed authority in place — narrower than the contract. Recording deferred pending a broadened seed-target match + enforcement binding (a dedicated hardening change, out of Track A recording scope)."
+  - validator_id: ontology-expansion-validator
+    obligation_id: validate_expansion_evidence_refs_against_valid_answer_support_ledger_or_seed_authority
+    coverage_status: pending
+    tier: flat
+    note: "slice-9 audit (2026-06-22): expansion evidence_refs ARE validated (unknown_id when not carried from cited answer claims; missing_required_ref when add/refine drops carried evidence), but against a PROXY — the cited answer claims' carried supporting_evidence_refs — not against the answer-support-ledger artifact or seed authority the name names. Neither is resolved in this validator. Partial/proxy; recording deferred pending direct ledger/seed-authority resolution or name reconciliation."
+  - validator_id: ontology-seed-validator
+    obligation_id: require_mixed_purpose_frame_projection_elements_to_preserve_member_source_and_cross_material_lineage_or_limitation
+    coverage_status: pending
+    tier: flat
+    note: "slice-26 audit (2026-06-23): NOT_FOUND in validateOntologySeed — the function never reads the named member-lineage fields (member_scope_refs / member_target_material_kind / member_source_refs / cross_material_ref_refs) on purpose_adequacy_frame.required_elements, and never discriminates on a `mixed` target. The mixed-target authority (target-material-profile-validation.yaml) is a registry-declared input_authority_ref but is NOT in this validator's function signature (declared≠wired). The analogous mixed-lineage enforcement is RECORDED on source-purpose-candidates-validator (require_mixed_purpose_frame_elements_to_carry_...). Recording deferred until the fields are read here."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_dynamic_boundaries_have_unique_ids_and_close_target_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-26 audit (2026-06-23): NOT_FOUND — validateOntologySeed never reads dynamic_layer.dynamic_boundaries[]; no uniqueness, boundary_kind enum, target_ref resolution, evidence_refs, or limitation_refs check exists for boundary rows. No bindable violation. Recording deferred pending implementation of dynamic-boundary validation."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_purpose_adequacy_frame_projection_equivalence_against_selected_validated_frame
+    coverage_status: pending
+    tier: activation_gated_dormant
+    note: "slice-26 audit (2026-06-23): activation_gated_dormant — gated on source_purpose_candidate_runtime_is_implemented; its conditional input (source-purpose-candidates-validation.yaml) is not received by validateOntologySeed's signature. No active enforcement; PARK until the source-purpose-candidate runtime is wired."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_required_element_closure_status_consistent_with_seed_evidence_and_limitation_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-26 audit (2026-06-23): PARTIAL — the validator enforces a PROXY (each required_element must cite seed_ref_refs OR limitation_refs, and those refs are resolved) but never reads the named `closure_status` field, so it cannot enforce the contract's status-specific consistency (modeled ⇒ seed+evidence; limitation_backed/frontier_required ⇒ limitation refs). Scope-discriminator (closure_status) unread → PARK."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_required_element_surface_dimension_closure
+    coverage_status: pending
+    tier: flat
+    note: "slice-26 audit (2026-06-23): NOT_FOUND — the required_elements loop reads element_id/element_kind/description/seed_ref_refs/limitation_refs/evidence_refs but never the named actionability_surface_refs[] or maturity_dimension_refs[] fields, so the surface/dimension closure enum-resolution the contract (operational-ontology-seed-contract §5.2) requires is unenforced. No bindable violation. PARK."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_seed_purpose_confirmation_projection_against_purpose_confirmation_validation
+    coverage_status: pending
+    tier: activation_gated_dormant
+    note: "slice-26 audit (2026-06-23): activation_gated_dormant — gated on purpose_confirmation_runtime_is_implemented; its conditional input (purpose-confirmation-validation.yaml) is not received by validateOntologySeed's signature. No active enforcement; PARK until the purpose-confirmation runtime is wired."
+  - validator_id: ontology-seed-validator
+    obligation_id: validate_seed_purpose_projection_equivalence_against_selected_source_purpose_candidate
+    coverage_status: pending
+    tier: activation_gated_dormant
+    note: "slice-26 audit (2026-06-23): activation_gated_dormant — gated on source_purpose_candidate_runtime_is_implemented; its conditional input (source-purpose-candidates-validation.yaml) is not received by validateOntologySeed's signature. No active enforcement; PARK until the source-purpose-candidate runtime is wired."
+  # slice-28 audit (2026-06-23; codex #143 R1): validateReconstructRunManifest delegates all 27
+  # governing_snapshot obligations to validateReconstructRunGoverningSnapshot, which rebuilds the
+  # EXPECTED snapshot from live authority and enforces each recorded snapshot field via ONE stableJson
+  # equality (subject_id = governing_snapshot.<field>; stamped at the checks loop, past the early-returns).
+  # RECORDED only the 4 reference-standard / pattern-catalog ID + VERSION obligations that map 1:1 to a
+  # NON-shared per-field equality whose authority-projection embodies the named scope (selected ids ==
+  # registry ids → resolve; version/snapshot map has an entry per selected id → recorded). The 23 below
+  # stay PARKED: most share a per-field equality with 2-6 sibling obligations (no isolated breaching
+  # binding), the "allowed/supported/contains-required-fields" constraints are enforced at snapshot-build
+  # time (the recorded-vs-rebuild equality passes if the authority itself omits a field), the canonical-URI
+  # obligation names a registry canonical_uri_policy the rebuild never reads (it uses a fixed URN
+  # convention; codex #143 R1 Finding 2), and a few named quantities are not compared snapshot fields at
+  # all (NOT_FOUND).
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: active_contract_hashes_match_active_contract_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). Enforced only via the single governing_snapshot.active_contracts stableJson equality (the rebuild computes each contract sha256 from its ref via sha256File), a field shared by obls active_contract_migration_status_values_are_allowed / active_contract_snapshots_contain_version_policy_required_fields / active_contract_snapshots_match_active_contract_refs_by_contract_id_and_ref — no isolated per-obligation breaching binding for the hashes-match-refs facet. DEFER to the registry input_authority_refs for the authoritative input set."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: active_contract_migration_status_values_are_allowed
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. The 'values are allowed' allow-list is enforced at snapshot-build time, not by the recorded-vs-rebuild equality (which only requires recorded.active_contracts to byte-equal the rebuild); the active_contracts field is also shared with obls 1/3/4 — not bindable in isolation."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: active_contract_snapshots_contain_version_policy_required_fields
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. 'contains version-policy required fields' is a build-time schema property; the equality passes if the rebuilt authority itself omits a field, so it does not independently enforce field presence. active_contracts field shared with obls 1/2/4."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: active_contract_snapshots_match_active_contract_refs_by_contract_id_and_ref
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). Enforced via the governing_snapshot.active_contracts equality shared with obls 1/2/3; no isolated breaching binding for the contract_id+ref match facet."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: admitted_domain_competency_snapshots_contain_version_policy_required_fields
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. 'contains version-policy required fields' is a build-time schema property (equality passes if the authority omits a field); the governing_snapshot.admitted_domain_competency_snapshots field is also shared with split_merged_replaced_competency_ids_close_against_admitted_domain_competency_snapshot."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: competency_id_migration_mappings_contain_required_policy_fields
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. Field-isolated on governing_snapshot.competency_id_migration_mappings, but 'contains required policy fields' is an internal-schema property guaranteed at snapshot build (buildAdmittedDomainSnapshots), not by the recorded-vs-rebuild equality, which passes if the authority itself omits a field."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: competency_id_migration_status_values_are_allowed
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. The competency-id migration-status allow-list is enforced at snapshot-build time, not by the recorded-vs-rebuild equality of the competency_id_migration_mappings field (which only requires byte-equality to the rebuild)."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: governed_seed_fields_close_against_selected_reference_and_pattern_refs
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (NOT_FOUND). Governed seed fields are not a governing_snapshot compared field; validateReconstructRunGoverningSnapshot performs no seed-vs-reference/pattern closure. The closure named here is not inspected by this validator."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: non_null_selected_source_profile_definition_refs_match_definition_sha256
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). The governing_snapshot.selected_source_profiles equality is shared by obls 9/12/23/24/25/26 (six source-profile obligations) and carries no isolated breaching binding for the definition_ref↔definition_sha256 match facet."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: p2_and_p3_competencies_remain_priority_metadata_until_explicit_downstream_promotion_policy_exists
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (AMBIGUOUS). Field-isolated on governing_snapshot.admitted_competency_priorities, but the equality only compares recorded priorities to the rebuild; it does not enforce the named non-promotion TEMPORAL policy (p2/p3 remain priority metadata until an explicit downstream promotion policy exists). No bindable violation for the policy clause."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: previous_competency_ids_resolve_to_current_id_or_explicit_replacement_disposition
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (NOT_FOUND). Previous competency ids and their current-id/replacement disposition are not a compared governing_snapshot field; the rebuild does not resolve previous→current ids, so the named resolution is not inspected here."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: previous_source_profile_snapshots_validate_old_to_current_profile_mappings
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). The old→current profile-mapping validation is not a distinct check; the governing_snapshot.selected_source_profiles equality (shared with obls 9/23/24/25/26) is the only enforcement, with no isolated binding."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: required_admitted_competency_ids_derive_all_p1_from_admitted_domain_competency_snapshot
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. The p1-derivation is a build property (the rebuild flattens required_admitted_competency_ids from the snapshots); the governing_snapshot.required_admitted_competency_ids equality only compares the resulting id set and is shared with required_admitted_competency_ids_match_admitted_domain_competency_snapshot — no isolated binding for the derive-all-p1 facet."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: required_admitted_competency_ids_match_admitted_domain_competency_snapshot
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). Enforced via the governing_snapshot.required_admitted_competency_ids equality shared with required_admitted_competency_ids_derive_all_p1_from_admitted_domain_competency_snapshot; no isolated per-obligation breaching binding."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_pattern_catalog_canonical_uris_satisfy_reference_pattern_catalog_registry_policy
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23, codex #143 R1 P2): PARK (name-broader-than-code, policy not read). The field-isolated governing_snapshot.selected_pattern_catalog_canonical_uris equality compares the recorded URIs against a rebuild that derives each URI from a FIXED URN convention of the pattern_catalog_ref_id (selectedPatternCatalogCanonicalUri) and never reads the registry's declared canonical_uri_policy. The registry policy (reconstruct-contract-registry.yaml: canonical_uri_policy=selected_run_manifest_value_required + canonical_uri_validation_rule) explicitly says the registry does NOT compare against a symbolic URI literal and requires the run-manifest-recorded value bound to ref_id+version_or_snapshot_id — the opposite of what the code enforces. So 'satisfy registry policy' is not enforced here; PARK until a policy-reading check exists."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_reference_and_pattern_migration_status_is_supported
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (NOT_FOUND). Reference/pattern migration status is not a distinct compared field — governing_snapshot.migration_status_values holds only source_profile + contract value sets — and 'is supported' is a build-time policy, not enforced by the recorded-vs-rebuild equality."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_registry_hash_matches_reconstruct_contract_registry
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (AMBIGUOUS, mode-dependent). The dedicated registry_sha256 early-return (governing-snapshot.ts ~836) is not isolatable: a wrong-hash mutation is also caught by the governing_snapshot.registry field equality, and the breach overlaps selected_registry_snapshot_contains_registry_ref_and_hash; in historical_replay mode the hash-match falls back to a presence-only shape check. No single isolated breaching binding."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_registry_snapshot_contains_registry_ref_and_hash
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. In live mode the registry ref+hash presence is enforced only via the governing_snapshot.registry field equality, which overlaps the hash-match early return (obl selected_registry_hash_matches_reconstruct_contract_registry); the dedicated presence shape check runs only in the historical_replay fallback. Not an isolatable live check."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_source_profile_migration_fields_are_recorded
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. 'migration fields are recorded' is a build-time/presence property; enforcement is the governing_snapshot.selected_source_profiles equality shared with obls 9/12/24/25/26 — no isolated binding."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_source_profile_snapshots_contain_version_policy_required_fields
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK. 'contains version-policy required fields' is a build-time schema property (equality passes if the authority omits a field); governing_snapshot.selected_source_profiles field shared with obls 9/12/23/25/26."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: selected_source_profile_snapshots_match_source_profile_records_by_profile_id_and_target_material_kind
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). Enforced via the governing_snapshot.selected_source_profiles equality shared with obls 9/12/23/24/26; no isolated breaching binding for the profile_id+target_material_kind match facet."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: source_profile_migration_refs_resolve_to_source_profile_records_or_prior_snapshots
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). The migration-ref resolution is not a distinct check; the governing_snapshot.selected_source_profiles equality (shared with obls 9/12/23/24/25) is the only enforcement, with no isolated binding."
+  - validator_id: pre-handoff-run-manifest-validator
+    obligation_id: split_merged_replaced_competency_ids_close_against_admitted_domain_competency_snapshot
+    coverage_status: pending
+    tier: flat
+    note: "slice-28 audit (2026-06-23): PARK (shared-field). Closure of split/merged/replaced competency ids is not a distinct check; the governing_snapshot.admitted_domain_competency_snapshots equality (shared with admitted_domain_competency_snapshots_contain_version_policy_required_fields) is the only enforcement, with no isolated binding."
+  - validator_id: purpose-confirmation-validator
+    obligation_id: require_revised_confirmation_to_preserve_source_conflict_or_trigger_purpose_discovery_rerun
+    coverage_status: pending
+    tier: flat
+    note: "slice-11 audit (2026-06-22): AMBIGUOUS — two-armed name, only one arm enforced. The rerun arm IS enforced (revised_pending_evidence_check → seed_readiness_effect must_rerun_purpose_discovery + conflicting_state, purpose-authority-validation.ts), but the 'preserve source conflict' arm is never checked: the validator never reads source_conflict_policy (or the confirmation's revised_statement vs source conflict) — only the rerun side-effect of revised_pending_evidence_check is realized. Recording deferred pending a source-conflict-preservation check."
+  - validator_id: purpose-confirmation-validator
+    obligation_id: validate_confirmation_status_against_source_purpose_candidate_status_and_validation_confirmation_required
+    coverage_status: pending
+    tier: flat
+    note: "slice-11 audit (2026-06-22): AMBIGUOUS — name broader than code. The validator reads sourceValidation.validation_status (conflicting_state when not valid) and sourceValidation.confirmation_required (the gate branch), but never the candidate's own purpose_source_status (explicit_source_declared/convergent_inferred/limitation_backed/unresolved) the name references. Per the contract confirmation_required is the only authoritative gate input, so the candidate-status arm is intentionally not read here — name names a quantity this validator does not inspect. Recording deferred pending name reconciliation (the confirmation_required arm is already recorded via require_confirmation_for_inferred_or_limitation_backed_purpose)."
+  - validator_id: reconstruct-run-control-validator
+    obligation_id: preserve_post_write_hash_observation_without_claiming_atomic_commit_when_writer_did_not_prove_atomic_rename
+    coverage_status: pending
+    tier: flat
+    note: "slice-10 audit (2026-06-22): NOT_FOUND in validateReconstructRunControl. The contract (design.md §run-control 'committed through an atomic write method or truthfully marked as observed_file_hash') governs commit_method, but the validator's write-transaction block (run-control-validation.ts) never reads commit_method — it checks only artifact_ref/owner_attempt_id presence and committed→committed_hash. Distinguishing atomic_rename from observed_file_hash and preventing an observed-hash transaction from claiming atomic commit is the writer's truthful-marking responsibility (recordReconstructRunControlTransactions), not validated. Recording deferred pending a validator-side commit_method check."
+  - validator_id: reconstruct-run-control-validator
+    obligation_id: reject_conflicting_request_fingerprints_before_semantic_artifacts_are_consumed
+    coverage_status: pending
+    tier: flat
+    note: "slice-10 audit (2026-06-22): AMBIGUOUS — name broader than the validator's check. The validator rejects request rows already MARKED duplicate_conflict/rejected_conflict (conflicting_request) and is the stage-0 gate consumed before semantic artifacts, but it never compares request_fingerprint values; fingerprint conflict DETECTION lives in the writer (initializeReconstructRunControl). The validator trusts the recorded request_status enum. Recording deferred pending a validator-side fingerprint comparison or name reconciliation."
+  - validator_id: reconstruct-run-control-validator
+    obligation_id: validate_current_attempt_and_session_root_lock_ownership
+    coverage_status: pending
+    tier: flat
+    note: "slice-10 audit (2026-06-22): AMBIGUOUS — name broader than code. The validator checks presence of an active attempt (active_attempt_missing), presence of a session_root lock (session_lock_missing), and rejects recorded conflicting locks (conflicting_lock), but it never cross-checks the lock row's owner_attempt_id against the current attempt's attempt_id. The named 'ownership' linkage is unverified — concurrency conflicts are trusted via the recorded lock_status enum. Recording deferred pending an owner_attempt_id linkage check."
+  - validator_id: reconstruct-run-control-validator
+    obligation_id: validate_session_root_request_fingerprint_target_signature_runtime_version_and_idempotency_are_replayable
+    coverage_status: pending
+    tier: flat
+    note: "slice-10 audit (2026-06-22): AMBIGUOUS — compound obligation, only 1 of 5 named quantities validated. Of session_root, request_fingerprint, target signature, runtime version, and idempotency key, the validator validates only session_root (session_root_missing presence + expected-root match); request_fingerprint, target_signature_ref, runtime_version, and idempotency_key_hash are never inspected, and validateReconstructRunControl's signature accepts no expected values for them. Recording deferred pending validation of the remaining four replayable quantities."
+  - validator_id: registry-verification-evidence-validator
+    obligation_id: validate_registry_snapshot_hash_matches_current_registry_file
+    coverage_status: pending
+    tier: flat
+    note: "slice-18 audit (2026-06-23): AMBIGUOUS — the hash-matches-current-file check (registry_hash_mismatch) is gated on the caller-supplied optional expectedRegistrySha256. validateRegistryVerificationEvidence never derives the on-disk registry hash itself, so it carries no internal 'matches current file' guarantee; the match is enforced only when the caller supplies sha256File(registryPath) (writeRegistryVerificationEvidenceValidationArtifact does so at the production call site, but the validator function does not). Recording deferred pending an internal hash-vs-file derivation. The validator's other 7 obligations are recorded (structural, unconditional). DEFER to the registry record's input_authority_refs for the authoritative input set."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_actor_action_state_scout_rows_do_not_replace_purpose_required_elements
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24): PARK (presence-only + material-source unchecked). The required boundary note 'Actor-action-state scout rows are prioritization evidence, not selected-purpose required elements.' is enforced presence-only (semantic_authority_boundary_missing). closure_row_dangling_required_element rejects closure rows whose required_element_ref is not a selected-purpose element, but never verifies a row's material is purpose-derived rather than scout-derived — a scout-sourced row labeled with a valid required_element_ref passes. 'do not replace' is not fully enforced."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_frontier_required_preserves_exploration_budget_state_without_declaring_source_insufficiency
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24): PARK (compound, only one half enforced). validateSeedAuthoringReadiness recompute-compares source_sufficiency_state (source_sufficiency_state_mismatch), so the 'without declaring source insufficiency' half is enforced; but exploration_budget_state is COPIED from the artifact into the validation output and never compared to the recomputed expected, so the 'preserves exploration_budget_state' half is unenforced. Recording deferred pending an exploration_budget_state comparison."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_only_seed_ready_or_limited_seed_possible_allows_seed_authoring
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24): PARK (DELEGATED). The allow/deny gate is enforced by the sibling assertSeedAuthoringReadinessAllowsSeed (throws unless readiness_classification is seed_ready or limited_seed_possible), NOT by validateSeedAuthoringReadiness — which returns validation_status=valid for frontier_required/blocked classifications (see test 'projects frontier_required without making validation invalid'). asserted_obligation_ids telemetry is on the validator's return, so the delegated gate cannot be recorded here."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_ontology_domain_required_category_rows_only_when_selected_purpose_closure_rows_resolve_to_domain_categories
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24): PARK (asymmetric 'only when'). The validator checks only the expected→actual direction (every recomputed ontology domain category id must be present, ontology_domain_category_missing). It never loops actual→expected to reject spurious/orphan category rows, so domain category rows present when the selected-purpose closure does not resolve to them are not rejected — the 'only when ... resolve to domain categories' direction is unenforced."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_readiness_consumes_pre_seed_source_scout_validation_snapshot
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24, codex #150 R1 P2): PARK (gated-on-optional-arg). The three source_scout_pre_seed_identity_mismatch checks (validation-ref basename + scout_pack_ref basename + concrete pre-seed sibling) are all conditional on args.sourceScoutPackValidationRef / args.sourceScoutPackValidation being present; an absent/null snapshot is handled by the validation-gap mechanism (validationGapSubjects → blocked_validation_gap), not the identity check. An unconditional top-of-function stamp would therefore over-claim pre-seed snapshot consumption for gap-only executions that never consume or verify a snapshot. Recording deferred pending a stamp gated on an actual snapshot-consumption check."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_readiness_projection_uses_validated_input_refs_only
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24): PARK (field never compared + shared mechanism). The closure-row validated_upstream_refs / llm_authority_refs projection fields are populated by the build but never compared in validateSeedAuthoringReadiness (only closure_state and material_admission_row_ref are). The only enforcement that gates on input validation status is the validation-gap mechanism, whose binding is readiness_classification_mismatch — already recorded for validate_blocked_validation_gap_is_projection_not_semantic_decision (no distinct binding for 'uses validated input refs only')."
+  - validator_id: seed-authoring-readiness-validator
+    obligation_id: validate_selected_purpose_required_elements_have_closure_rows
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-4 audit (2026-06-24, codex #150 R1 P2): PARK (duplicate-element_id Map collapse). validateSeedAuthoringReadiness keys both expected and actual closure rows by required_element_ref in a Map; if a selected purpose frame contains two required elements with the same element_id (source-purpose validation enforces non-empty element_id but does NOT reject duplicates), the duplicates collapse and a single closure row satisfies both, so closure_row_missing is never raised for the omitted duplicate. 'every selected-purpose required element has a closure row' is not fully enforced. Recording deferred pending duplicate-element_id rejection or per-occurrence coverage."
+  - validator_id: source-observation-delta-validator
+    obligation_id: validate_delta_frontier_kind_is_supported
+    coverage_status: pending
+    tier: flat
+    note: "slice-5 audit (2026-06-22): normalizeFrontierForDelta validates kind-vs-artifact CONSISTENCY (frontier_kind_mismatch when the frontier artifact does not match the declared source_frontier / maturation_closure_frontier kind), but does NOT validate that the frontier_kind value itself is in a supported enumeration — an unsupported kind branches to the maturation_closure path without an explicit unsupported-kind violation. Recording deferred pending an explicit supported-kind check or name reconciliation."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: require_each_lineage_row_delta_validation_to_be_valid
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23, codex #149 R1 P2): PARK (null-read false-pass). A delta-validation ref pointing at an EXISTING EMPTY/null YAML makes readYamlDocument return null WITHOUT throwing, so the catch (lineage_delta_validation_missing, ~728-734) never fires AND the validity check (~769: `if (deltaValidation && status!=='valid')`) is skipped — the row passes valid with a non-valid/absent delta validation. Not fully enforced until non-object reads are rejected."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: require_each_lineage_row_reentry_validation_to_be_valid
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23, codex #149 R1 P2): PARK (null-read false-pass). Same as the delta-validation obligation: an existing empty/null re-entry-validation YAML returns null without throwing, so lineage_reentry_validation_missing (~742-747) never fires and the `if (reentryValidation && status!=='valid')` check (~817) is skipped. Not fully enforced until non-object reads are rejected."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: validate_each_lineage_added_observation_exists_in_source_observations
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23, codex #149 R1 P2): PARK (caller-supplied authority). The check (~884-890 → lineage_observation_missing) verifies each added id is present in the PASSED sourceObservations object, but never binds that object/ref or session to the delta/re-entry chain's source_observations_ref. A wrong or stale source-observations file containing the same ids therefore passes and is trusted downstream as the validated observations authority. Not fully enforced until the observations authority is bound to the chain."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: validate_each_lineage_added_observation_was_reentered_by_its_validation
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23): PARK (shared violation code). The reentered-membership check (~891-897: reentryValidation && !reentered.has(observationId)) emits the SAME lineage_added_observation_mismatch code as validate_lineage_added_observation_ids_match_delta_added_observation_ids (set-equality, ~862-867). They fire from distinct sites but share one code, so this obligation has no distinct breaching binding (slice-21). It is also subject to the null-read false-pass (an empty re-entry-validation file skips the check). Record deferred pending a dedicated reentered-mismatch code and non-object-read rejection."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: validate_each_lineage_row_delta_ref_is_readable_and_session_matching
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23, codex #149 R1 P2): PARK (null-read false-pass). 'readable' is not enforced: a delta ref pointing at an existing empty/null YAML returns null without throwing, so lineage_delta_missing (~716-720) never fires and the session/round/frontier matches (~748-767, all gated on `if (delta && ...)`) are skipped — an empty-but-present delta file passes. Not fully enforced until non-object reads are rejected."
+  - validator_id: source-observation-lineage-index-validator
+    obligation_id: validate_lineage_added_observation_ids_match_delta_added_observation_ids
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-3 audit (2026-06-23, codex #149 R1 P2): PARK (null-read false-pass). The set-equality match (~862-867 → lineage_added_observation_mismatch) is gated on `if (delta)`; an existing empty/null delta YAML returns null without throwing, so the match is skipped and arbitrary added_observation_ids pass. Not fully enforced until non-object reads are rejected."
+  - validator_id: source-purpose-candidates-validator
+    obligation_id: derive_confirmation_required_from_selected_purpose_source_status
+    coverage_status: pending
+    tier: flat
+    note: "slice-24 audit (2026-06-23): PROJECTION-ONLY. The validator derives confirmation_required (~420-423: selected.purpose_source_status != explicit_source_declared) and emits it in the return object (~434); there is no violation and no breaching input that trips one, so there is no non-vacuous enforcement binding to record. Contract design.md:405 frames it as a derived runtime predicate field, not a check. Recording deferred pending a downstream consumer-validator that gates on it (cf. purpose-confirmation slice-11, which records consuming confirmation_required)."
+  - validator_id: source-purpose-candidates-validator
+    obligation_id: expose_selected_purpose_candidate_id_and_frame_id_for_seed_projection_equivalence
+    coverage_status: pending
+    tier: flat
+    note: "slice-24 audit (2026-06-23): PROJECTION-ONLY. selected_purpose_candidate_id and selected_purpose_frame_id are computed and emitted in the return object (~420,432-433) with no violation; nothing a breaching input trips. Contract design.md:408 asks that they be 'explicit so seed validation can prove projection equivalence' — the equivalence check belongs to seed validation, not here. Recording deferred to the seed-validation consumer."
+  - validator_id: source-purpose-candidates-validator
+    obligation_id: require_exactly_one_primary_purpose_candidate_when_selected
+    coverage_status: pending
+    tier: flat
+    note: "slice-24 audit (2026-06-23): superset-blind-to-discriminator (slice-20 pattern). The code enforces exactly-one-primary UNCONDITIONALLY (~173-186: missing_primary_purpose if 0, multiple_primary_purpose if >1), but the contract (design.md:390) scopes the rule to 'exactly one primary candidate WHEN THE PURPOSE IS SELECTED'. The validator never reads the selected/not-selected discriminator and has no dedicated when-selected violation code, so it validates a broader unconditional rule rather than the obligation's named scope. Recording deferred pending a selected-scoped enforcer (or contract re-scoping to unconditional)."
+  - validator_id: source-purpose-candidates-validator
+    obligation_id: require_purpose_candidate_status_field_to_be_purpose_source_status
+    coverage_status: pending
+    tier: flat
+    note: "slice-24 audit (2026-06-23): PARTIAL. The only enforcement is the artifact-level alias-field rejection (~153-160, alias_field_present: rejects source_purpose_status/inference_status keys). There is no per-candidate check that candidate.purpose_source_status is a valid enum value (PURPOSE_SOURCE_STATUSES), which the obligation name ('candidate status field to be purpose_source_status') implies. Recording deferred pending per-candidate purpose_source_status enum validation."
+  - validator_id: source-safety-ledger-validator
+    obligation_id: preserve_prompt_context_evidence_support_public_output_replay_and_material_claim_consumption_boundaries
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-1 audit (2026-06-23): PARK (overlapping). The only enforcement is the per-row intended_consumption enum (invalid_enum, ~520-526) plus the per-observation-per-consumption required-row presence — and that presence check is itself a weak proxy (see validate_every_observation...). There is no independent check that one consumption row may not substitute for another's authority, so this obligation cannot bind a non-overlapping breaching input. Record deferred pending a dedicated no-substitution enforcer."
+  - validator_id: source-safety-ledger-validator
+    obligation_id: validate_every_observation_has_source_safety_rows_for_each_intended_consumption
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-1 audit (2026-06-23, codex #147 P1): PARK (weak proxy). The required-row check (requiredSafetyRowIds per observation × INTENDED_CONSUMPTIONS, ~414-420/541-549 → source_observation_safety_row_missing) only proves each source_safety:<observation_id>:<intended_consumption> ID STRING is present; it never binds that ID's consumption suffix to the row's visibility_derivation.intended_consumption — the field deriveSourceSafetyVisibilityTier actually consumes. So with every required ID present, a public_output/material_claim-ID row can carry prompt_context as its derivation consumption and a derived consumption_allowed tier, pass this validator, and mislead downstream consumers that look up the tier by row ID. Record deferred pending an ID-suffix to intended_consumption binding."
+  - validator_id: source-scout-pack-post-maturation-validator
+    obligation_id: validate_post_maturation_scout_snapshot_is_immutable_audit_authority
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (immutability/audit-authority not enforced). The shared validateSourceScoutPack checks INPUT-artifact snapshot hashes (source-observations/safety match current files = freshness) but never pins the snapshot's own content hash nor checks any audit-authority state, so a mutated post-maturation snapshot re-hashed to current inputs passes."
+  - validator_id: source-scout-pack-post-maturation-validator
+    obligation_id: validate_prompt_visible_rows_have_source_safety_validation_refs
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (named per-row field unread). The named per-row field source_safety_ledger_validation_ref is built but never read by validateSourceScoutPack; it checks source_safety_row_id (ledger row) + global validation_status + consumption_allowed tier instead. A prompt-visible row with a valid source_safety_row_id but null/wrong source_safety_ledger_validation_ref passes → the named validation-ref field is unenforced."
+  - validator_id: source-scout-pack-post-maturation-validator
+    obligation_id: validate_snapshot_lineage_validation_ref_and_hash_match_current_lineage_authority
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (slice-3 null-read + caller-authority). source_observation_lineage_index_validation_ref/hash match is checked, but sha256File(null)=null so a null lineage ref + null pack snapshot hash passes without a violation, and the 'current lineage authority' is the caller-supplied ref (not derived from the chain). Lineage validation is a required pre/post input, so the null-pass is a false-pass."
+  - validator_id: source-scout-pack-post-maturation-validator
+    obligation_id: validate_snapshot_scope_derives_from_target_material_profile_validation
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (not derived from validation). scout_scope is copied from the pack; validateSourceScoutPack only checks internal consistency (unsupported_scope_overclaimed: scope_state vs the scout's own target_ref_count/material_kind) and the validation's validation_status=valid. It never recomputes/derives scout_scope from the target-material-profile-validation content (scoutScopeState() exists but is unused in the validator)."
+  - validator_id: source-scout-pack-pre-seed-validator
+    obligation_id: validate_pre_seed_scout_snapshot_is_immutable_consumed_authority
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (immutability/consumed-authority not enforced). The shared validateSourceScoutPack checks INPUT-artifact snapshot hashes (source-observations/safety match current files = freshness) but never pins the snapshot's own content hash nor checks any consumed-authority state, so a mutated pre-seed snapshot re-hashed to current inputs passes."
+  - validator_id: source-scout-pack-pre-seed-validator
+    obligation_id: validate_prompt_visible_rows_have_source_safety_validation_refs
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (named per-row field unread). The named per-row field source_safety_ledger_validation_ref is built but never read by validateSourceScoutPack; it checks source_safety_row_id (ledger row) + global validation_status + consumption_allowed tier instead. A prompt-visible row with a valid source_safety_row_id but null/wrong source_safety_ledger_validation_ref passes → the named validation-ref field is unenforced."
+  - validator_id: source-scout-pack-pre-seed-validator
+    obligation_id: validate_snapshot_lineage_validation_ref_and_hash_match_current_lineage_authority
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (slice-3 null-read + caller-authority). source_observation_lineage_index_validation_ref/hash match is checked, but sha256File(null)=null so a null lineage ref + null pack snapshot hash passes without a violation, and the 'current lineage authority' is the caller-supplied ref (not derived from the chain). Lineage validation is a required pre/post input, so the null-pass is a false-pass."
+  - validator_id: source-scout-pack-pre-seed-validator
+    obligation_id: validate_snapshot_scope_derives_from_target_material_profile_validation
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (not derived from validation). scout_scope is copied from the pack; validateSourceScoutPack only checks internal consistency (unsupported_scope_overclaimed) and the validation's validation_status=valid. It never recomputes/derives scout_scope from the target-material-profile-validation content (scoutScopeState() exists but is unused in the validator)."
+  - validator_id: source-scout-pack-validator
+    obligation_id: validate_group_and_coverage_refs_resolve
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (compound, 'group' half has no enforcement surface). Only coverage_slot→signal_row resolution (coverage_slot_signal_missing) is checked; there is no 'group' structure in the scout-pack artifact (the word appears nowhere in the validator), so the 'group ... refs resolve' half is unenforced."
+  - validator_id: source-scout-pack-validator
+    obligation_id: validate_prompt_visible_rows_have_source_safety_validation_refs
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (named per-row field unread). The named per-row field source_safety_ledger_validation_ref is built but never read by validateSourceScoutPack; it checks source_safety_row_id (ledger row) + global validation_status + consumption_allowed tier instead. A prompt-visible row with a valid source_safety_row_id but null/wrong source_safety_ledger_validation_ref passes → the named validation-ref field is unenforced."
+  - validator_id: source-scout-pack-validator
+    obligation_id: validate_scout_pack_is_profile_local_and_contains_no_selected_purpose_required_element_refs
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (compound, 'profile-local' half unchecked + substring proxy). validateNoSelectedPurposeLeak is a JSON.stringify().includes('purpose_required_element_ref') substring search (no-purpose-leak half, fragile to field-name variants); the 'profile-local' half is never checked by the validator."
+  - validator_id: source-scout-pack-validator
+    obligation_id: validate_scout_scope_derives_from_target_material_profile_validation
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-5 audit (2026-06-24): PARK (not derived from validation). scout_scope is copied from the pack; validateSourceScoutPack only checks internal consistency (unsupported_scope_overclaimed) and the validation's validation_status=valid. It never recomputes/derives scout_scope from the target-material-profile-validation content (scoutScopeState() exists but is unused in the validator)."
+  - validator_id: target-material-profile-validator
+    obligation_id: mixed_targets_preserve_member_to_aggregate_purpose_requirement_mapping_or_limitation
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-2 audit (2026-06-23): PARK (NOT_FOUND). The mixed branch (~154-178) only checks that the selected profiles include the registry mixed row and one per candidate kind (mixed_candidate_profile_missing); there is no member→aggregate purpose-requirement mapping or limitation check, and ReconstructSelectedSourceProfileRef carries no purpose_requirement / purpose_mapping / limitation_refs field. The named mapping is not inspected here."
+  - validator_id: target-material-profile-validator
+    obligation_id: mixed_targets_preserve_per_member_selected_profile_id_contract_status_runtime_implementation_status_and_support_state
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-2 audit (2026-06-23): PARK (PARTIAL, scope + missing facet). contract_status and runtime_implementation_status ARE match-checked against the registry (~241-246 → selected_profile_registry_mismatch), but that check runs for ALL selected profiles, not the named mixed_targets/per_member scope (the validator never gates this on target_material_kind=='mixed'), and the named support_state facet is not a field on ReconstructSelectedSourceProfileRef / ReconstructSourceProfileRecord at all. The mixed-scoped per-member preservation is not the enforced quantity (slice-20 scope-discriminator pattern)."
+  - validator_id: target-material-profile-validator
+    obligation_id: mixed_targets_preserve_per_member_source_profile_snapshot_ref_definition_hash_and_source_refs
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-2 audit (2026-06-23): PARK (PARTIAL, scope + missing facet). definition_sha256 and profile_ref ARE match-checked against the registry (~241-246), but unscoped (all selected profiles, not the named mixed scope), and the named source_refs facet is not a field on the selected-profile snapshot (scan_targets exists but is neither source_refs nor a required/matched field). The mixed-scoped per-member preservation is not the enforced quantity."
+  - validator_id: target-material-profile-validator
+    obligation_id: partial_composite_member_support_limits_aggregate_ready_projection
+    coverage_status: pending
+    tier: flat
+    note: "deferred-7 slice-2 audit (2026-06-23): PARK (NOT_FOUND). support_status is an aggregate field on the target-material-profile artifact, not on the selected members; the validator checks only that a supported/partial/composite aggregate selects ≥1 profile (~180-193 → selected_profile_missing). There is no check that a partial_composite MEMBER's support state downgrades/limits the aggregate ready projection. The member→aggregate support-limiting rule is not inspected here."