XRP Risk
Privacy-preserving credit scoring for decentralized lending on XRPL. Prove creditworthiness via a trusted enclave, receive a verifiable on-chain grade.
Project Information
At a Glance
Privacy-preserving credit scoring for decentralized lending on XRPL. Prove creditworthiness via a trusted enclave, receive a verifiable on-chain grade.
Description
XRP Risk is the third path between overcollateralization and centralized identity exposure. A borrower connects their XRPL wallet, links bank data through a Trusted Execution Environment where raw financial data never leaves the enclave, and receives a verifiable credit grade issued as an XLS-70d credential on the XRP Ledger.
The grade is signed as a W3C DID-JWT, portable and cryptographically verifiable by anyone.
The scoring engine blends on-chain behavior (70%) with off-chain financials (30%) across six weighted signals including payment punctuality, counterparty diversity, and cycle integrity. A self-dealing detector flags round-trip payments within 72 hours and zero-weights them. Score decay penalizes dormancy at 2% per week.
Grade A borrowers access loans at 5% APR. Grade B at 8.5%. First loan capped at 50 XRP, scaling to 1,000 XRP after three successful repayments. Walking away destroys access to progressively better terms across the entire protocol.
Technical Details
XRP Risk is built on Node.js with a Vite and React frontend in TypeScript. The backend runs Express with SQLite in WAL mode for fast demo queries, with production-ready code targeting XRPL Lending-Devnet.
The protocol uses three native XRPL primitives: XLS-65 Single Asset Vaults for liquidity with LP share tracking, XLS-66 Loan Broker with First-Loss Capital funded by the spread between borrower and vault rates, and XLS-80 Permissioned Domains for credential-gated vault access. Hook middleware verifies XLS-70d credentials cryptographically and enforces graduated loan caps before any disbursement.
The simulated TEE blends behavioral scores with mock Plaid financials, signs the grade as an EdDSA DID-JWT, and issues a CredentialCreate on XRPL Devnet. An AI Risk Agent powered by Claude provides data-grounded portfolio analysis with full context injection.
A forensic force-directed network graph visualizes counterparty relationships and flags AML-risk nodes.
Team
2Nikola Nikolov
Anoushk kv
Hackathon
Stablecoins & Payments Hackathon
Duration
Mar 21, 8:00 AM - Mar 22, 4:00 PM UTC
Prizes Won
1Prizes Won
1 prize awarded