Status: Phase 1 placeholder. Substantive content lands in Phase 5 per §8.1. Note: the V- prefix is shared with V-2 (voice & tone). The collision is intentional in the roadmap — V-1 = Vault, V-2 = Voice.

Statement

Vault (the encrypted personal drive) is private storage for files the user owns. Files are end-to-end encrypted with per-file keys, uploaded to Blossom servers, and indexed via kind:30078 metadata events on the user's relays. The UI must respect that the metadata events are the source of truth — Blossom is replaceable storage, the relays are the registry.

Why

Vault is the surface most easily mistaken for a generic cloud-drive. That misreading produces UI choices that fail privately but visibly: thumbnail previews fetched from third-party CDNs, search that hits a server, "shared with me" surfaces that imply a shared folder model the protocol doesn't support.

Vault's design discipline is: the file is the primary object; folders are tags; sharing is link-based; previews are local. Every UI affordance flows from those four facts.

How

This stub is intentionally thin. Substantive V-1 content matures in Phase 5. Expected coverage at maturity:

File-list as canonical surface; per-file preview as secondary
Folder navigation as tag-filter, not a hierarchical tree (UI may render as tree, but the model is flat-with-paths)
Share-link creation as the explicit privacy-relevant moment — password gating, expiration, revocation
Trash with restore and permanent-delete affordances
Storage stats: total size, blob cache size, recently accessed
Empty states (P-3) per folder

Phase 5 fills these in with audit citations and per-surface specifications.