Referral programs live or die on members understanding their network, but genealogy UIs are usually bespoke: one tree for unilevel, another for binary spillover, each hand-wired to a backend. And the moment an agent should read or drive the view, a bespoke tree means DOM scraping.

DownlineTree is a controlled genealogy surface: a flat, JSON-friendly member list in (each row carrying BOTH sponsorId and placementId), a collapsible tree out. The edge prop picks which pointer draws the shape — "sponsor" (unilevel) or "placement" (binary / matrix) — so one dataset renders every downline type. Controlled selection, tier Badges via tierColor, dimmed inactive members, stable data-mlm-node handles.

Feed it the member rows your particle-academy/fancy-mlm (or @particle-academy/fancy-mlm) backend already stores, set edge from your plan's tree type, and wire selectedId + onSelect. The sandbox's /referrals page is exactly this over the live engine.