Patient-specific computational models of atrial electrophysiology (EP) are increasingly important for understanding pathological conditions like atrial fibrillation (AF). A key objective is to generate anatomical models that accurately replicate clinical data across diverse cases. Models are typically constructed using electroanatomical maps or imaging data like computed tomography (CT) and magnetic resonance imaging (MRI), but current frameworks often require manual inputs and are constrained to surface or bilayer geometries. In this study, we introduce an efficient, fully automated computational framework for generating personalized, volumetric atrial models. Building on the AugmentA pipeline, our approach starts with segmented CT or MRI data and automatically extrudes the endocardial and epicardial walls of the left and right atrium with prescribed wall thicknesses. A comprehensive set of the most important anatomical regions for EP modeling and the corresponding fibers are then included by means of a rule-based approach. Additionally, a new method to represent interatrial connections is represented, enhancing model flexibility while preserving the physiological electrical connections between the atria. The framework also introduces a universal reference system, Universal Atrial Coordinates, for spatial parameter alignment, and an automated fitting process to generate a realistic and ready-to-use torso anatomy around the atria to simulate realistic electrocardiograms (ECGs). The reliability and scalability of our framework are demonstrated by testing our approach on a cohort of 50 AF patients. We moreover show that our frameworks support automatic swipping of EP parameters, enabling patient-specific calibration of the EP model based on ECG data. Our results show promise for applications in personalized treatment planning and research into atrial pathological conditions.