The NOAA/Global Systems Laboratory and the National Center for Atmospheric Research have complementary Hurricane Supplemental projects to explore the addition of community tools to the Hurricane Analysis and Forecast System (HAFS) workflow. HAFS is expected to be used by NOAA for operational tropical cyclone forecasts in the future, and is a key application in NOAA's Unified Forecast System. One of the tools currently under development for HAFS is the Community Data Models for Earth Prediction Systems (CDEPS). CDEPS can be used to run atmosphere-ocean coupled experiments with offline datasets, including those from field campaigns, reanalyses, and model forecasts. By using these canned and/or idealized datasets to initialize and force a component of the modeling system, researchers can selectively disable some of the feedbacks within the coupled system. This hierarchical testing approach reduces the number of nonlinear interactions in the model forecasts, making it easier to isolate and understand phenomena of interest. This presentation will discuss the implementation of CDEPS in HAFS and show some examples of tropical cyclone forecasts made with CDEPS.