Every RFC and active Internet-Draft in the corpus. Filtered client-side from a single JSON dataset — no server query.