Most backups are only tested on disaster day. SHORE cuts a manifest at the source — a fingerprint per table/file (count + content hash with a stable order) — and proves, on restore, that what came back is identical to what left.
Not a yes/no: a proven ladder — loads → consistent → matches source → point-in-time → business invariants. Each rung is a proof, not an assumption.
An untested backup is a promise; SHORE is a proof. For anyone with a continuity obligation (bank, health, regulated), the gap between 'we think it restores' and 'we prove it restores' is the product.
Step-0 measured on a Postgres toy DB (level 1/3/5, two attacks caught by the hash alone). It composes with GIRDER: the restore proof becomes a signed ledger entry.