Enables hybrid search (dense + sparse) over self-hosted indexes of continuously ingested public datasets (news, GitHub, Wikipedia, arXiv, small web, devdocs, Hacker News) using your own embedding model, served via the Model Context Protocol.
