MLKY

Graded trading cards are a large asset class with almost no liquidity. Lending against them is difficult for reasons that have little to do with crypto. A physical, professionally graded object has to be bound to a digital one, and that binding has to stay honest. The collateral has to be priced without a continuous market. And the loan has to settle atomically.

The protocol divides responsibility along the trust boundary. Custody, pool exposure and settlement are enforced on-chain, where they cannot depend on anyone's good behaviour. Valuation and underwriting sit in a service that signs each loan and withholds credit when market data degrades. Collateral eligibility is proven cryptographically at the point of borrowing, which is what makes double-pledging impossible under either the asset's digital or physical identity.

Two areas carry most of the risk in a protocol like this, and both were engineered accordingly. The first is the boundary where the protocol decides whether to trust a valuation at all: we identified and closed a signature-verification gap there that would otherwise have let one legitimate quote authorise arbitrary ones. The second is schema evolution, where the protocol now enforces a tested allocation invariant across every account type, so growth in one record cannot orphan positions already on chain. Neither is visible in the product. Both are preconditions for holding someone's collection.

Default resolution is automatic. A defaulted loan sends its collateral to auction, bids are held in escrow, and proceeds settle in a fixed order with a reserve fund behind them. The protocol has been scoped for external audit against a tagged release candidate, and runs today as an invite-only beta.