cmd/core-runtime/cognitive_interpretation.go

@@ -75,7 +75,13 @@ func shouldRequestSemanticInterpretation(result interactiondomain.SemanticResult
 	if result.Interpretation.Clarification == interaction.ClarificationRequired {
 		return false
 	}
-	if result.NeedsCognition || result.InformationRequestKind != "" || len(result.CapabilityHints) > 0 {
+	// Structured, read-only information intents already have a semantic owner
+	// in core-runtime. Sending them through the slower advisory interpretation
+	// path only adds latency and can make a healthy runtime look unavailable.
+	if result.InformationRequestKind != "" {
+		return false
+	}
+	if result.NeedsCognition || len(result.CapabilityHints) > 0 {
 		return true
 	}
 	if result.Interpretation.Applicability != interaction.SemanticApplicable {

cmd/core-runtime/cognitive_interpretation_test.go

@@ -36,7 +36,7 @@ func TestOperationalQuestionStillUsesStrictInterpretation(t *testing.T) {
 	}
 }
 
-func TestDirectiveAndInventoryPreferMeaningInterpretationEvenWhenDeterministicLooksApplicable(t *testing.T) {
+func TestStructuredInventoryUsesDirectReadOnlySemanticRoute(t *testing.T) {
 	result := interactiondomain.SemanticResult{
 		Interpretation: interaction.UtteranceInterpretation{
 			SpeechAct:        interaction.SpeechActDirective,
@@ -45,8 +45,8 @@ func TestDirectiveAndInventoryPreferMeaningInterpretationEvenWhenDeterministicLo
 		},
 		InformationRequestKind: interaction.InformationRequestCognitiveModelInventory,
 	}
-	if !shouldRequestSemanticInterpretation(result) {
-		t.Fatal("model inventory directive must be interpreted by meaning, not only token reflex")
+	if shouldRequestSemanticInterpretation(result) {
+		t.Fatal("structured model inventory must use the direct read-only semantic route")
 	}
 }