Enables searching ArXiv, fetching and indexing papers, and querying a personal paper library using vectorless RAG with BM25 and contextual compression, all without GPU or embedding models.
