01Semantic chunking to keep multi-page sections together, ensuring contextual integrity.
02Automatic extraction of requirements, hex values, tables, and figures from PDFs.
03Multiple search modes: BM25 (keyword), semantic (meaning), and hybrid (default).
04Persistent index with smart caching for fast subsequent searches and re-indexing checks.
051 GitHub stars
06Hybrid search combining BM25 keyword and semantic embeddings with RRF fusion.