Identifying the hotspots for damage initiation in large-scale composite structure designs presents a significant challenge due to the high modelling cost. For most industrial applications, the finite element (FE) models are often coarsely meshed with shell elements and used to predict the global stiffness and internal loads. Because of the lack of detailed descriptions for the composite materials and 3D stress states, most of the established failure criteria are not applicable. In this work we present an Abaqus plugin tool which implements a framework to identify the hotspots by using a pre-computed database generated for specific, heavily-repeated feature types based on a given structural model. Developed with an object-oriented implementation in Python, this software is split into two main parts, specifically for feature generation and structural analysis. The pre-computed model presents a full 3D description for the considered feature and works as a submodel to the coarse structure model driven by a one-way transfer of the boundary conditions. The presented framework is an analysis tool for efficient sizing of large-scale composite structures, as it enables 3D damage analysis of the structures in critical zones with significant savings of the modelling and computational cost. The results are compared with conventional FE modelling and satisfactory agreement is observed. In addition, the software also enables the pre-computed database to be stored in an HDF5 data file for further reuse on new structures with the same feature.