abrenunt.io is a project to make Catholic primary sources accessible to AI assistants through the Model Context Protocol.
The collection focuses on public-domain texts: papal encyclicals, writings of the Church Fathers, conciliar documents, and theological treatises. Documents are processed into structured, searchable chunks using hybrid semantic and keyword search (dense embeddings + BM25).
Source documents (PDFs, HTML, EPUB) are processed into structured blocks, chunked by section, and embedded using a multilingual sentence transformer model. Search queries are matched using Reciprocal Rank Fusion of dense semantic similarity and sparse BM25 keyword matching, enabling both conceptual search and exact term lookup in over 30 languages with language-specific stemming.
For questions, suggestions, requesting document not yet in the store, or to report issues: