Data model (Firestore mirror)

TL;DR

• Each vault is mirrored into Firestore for fast UI reads.
• Factory ID maps to collection name; vault account ID maps to document ID.
• Indexed fields drive Discover, vault views, and positions.

Collection layout

• Collection key: factory ID (nzaza.testnet or sudostake.near).
• Document key: vault account ID (for example vault-12.nzaza.testnet).

Core fields used by UI

• owner
• state (idle, pending, active) derived from liquidity request + accepted offer
• liquidity_request (token, amount, interest, collateral, duration)
• accepted_offer (lender, accepted_at)
• liquidation
• unstake_entries
• current_epoch
• indexing metadata (factory_id, tx_hash, created_at, updated_at)

Refresh model

• On key mutations, the app re-reads on-chain vault state and rewrites the document.
• If data appears stale, retry indexing from the UI.