Collected molecules will appear here. Add from search or explore.
SearchQuery -> RankedList<DocChunk>
Combine BM25 full-text queries via SQLite FTS5 with cosine similarity scores from sqlite-vec to retrieve and rank matching document chunks.
Problem it solves
Pure vector search misses precise keyword/API signature matches, while pure lexical search misses semantic context.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.