the limitations currently are storing the embedding of those very large codebases i would say because it takes time to indexing such large repo