SHARE (Spanish HARmful Expressions) is a new lexical resource composed of insults and offensive expressions collected using the Fiero chatbot and then manually labeled by 5 annotators. We used SHARE to release OffendES_spans which is the OffendES corpus automatically annotated with offensive entities relying on SHARE. We explore the usefulness of SHARE for the interpretability of offensive comments by comparing it with a BERT-based fine-tuning model.