Complex clinical decision support (CDS) that goes beyond representing simple clinical flowcharts to supporting the totality of a care encounter may help improve care quality and consistency. However, integrating a large volume of clinical guidelines applicable to a care encounter poses unique design and safety considerations. We present the visual and technical methods employed in developing NoviGuide, a platform for complex CDS. Assuring safe functioning required transparency of all outputs, which we achieved using a JSON formalism for capturing logic. Unlike raw computer code, logic-as-data can be presented clearly in context to non-informatician reviewers. Two different styles for visualizing CDS logic, random-access and narrative, support different review contexts. We assess the fitness of these solutions for encoding hundreds of neonatal-care guidelines into integrated multi-topic CDS.