The increasing complexity of cybercrime presents significant challenges for prosecutors in evaluating the readiness and success probability of criminal prosecution. Conventional approaches primarily rely on legal interpretation and professional judgment, often overlooking the interdependencies among evidentiary factors that influence prosecutorial outcomes. This study aims to develop a prosecution risk architecture for cybercrime cases using a Bayesian Network approach to model probabilistic relationships among key prosecution variables. A quantitative probabilistic modeling approach was employed by constructing a Bayesian Network consisting of evidence quality, chain of custody integrity, seizure procedure validity, digital forensic competence, offender identification, cross-jurisdictional cooperation, prosecutorial indictment quality, and prosecution success probability. Simulation results demonstrate that electronic evidence quality and chain of custody integrity are the most influential variables affecting prosecution success. Furthermore, the proposed model effectively captures risk propagation among interconnected variables, enabling dynamic evaluation of prosecution scenarios under uncertain conditions. The developed architecture provides a transparent and evidence-based decision support framework for assessing prosecution readiness and identifying critical factors requiring improvement before trial. This study contributes to the advancement of risk-based prosecution by introducing a probabilistic analytical framework capable of supporting more objective, adaptive, and systematic prosecutorial decision-making in addressing increasingly sophisticated cybercrime cases.