VaultScore reads public ledger data and computes a transparent, reproducible 300–850 score for any XRPL wallet — the same way a FICO score works, except you can check the math yourself. No login. No wallet connection. Nothing ever submitted on-chain just to get scored.
XRPL's native lending protocol (XLS-66) is coming, but lending needs a risk layer to mean anything — some way to tell a wallet with years of disciplined on-chain behavior apart from one created yesterday. Right now, there isn't one. Every lender re-derives trust from scratch, every time, for every counterparty.
THIN_FILE. VaultScore still scores everything else that is observable today (account age, activity depth, credentials), so the network isn't starting from zero the moment lending goes live.
VaultScore's Layer 1 model is fully deterministic — every input is something anyone can recompute by hand from public ledger data, with fixed arithmetic and no hidden model. That's what makes disputes fast instead of a 30-day black box, and it's why every score ships with plain-language reason codes explaining exactly what moved it.
Read the full methodology →No connection, no signature, no approval popup — just the public r-address you want a read on.
Account info, transaction history, ledger objects and trust lines are pulled live from XRPL mainnet.
Six weighted factors combine into a 300–850 score, with reason codes explaining exactly why.
Opt in and VaultScore re-checks the score every month on its anniversary, building a real trend line.
Layer 1 scoring uses fixed arithmetic on public data — no ML, no hidden weights, nothing to trust blindly.
Every score is computed live against XRPL mainnet — not a cached demo, not synthetic test data.
Every score explains its top drivers — the same function as a bureau's adverse-action notice, offered up front instead of on request.
Scores can be packaged into an XLS-70 Credential and issued directly to the wallet — independently verifiable by anyone.
Opt any wallet into a running history, re-checked automatically every month on its own anniversary.
The same scoring core ships as an MCP tool — Claude, Cursor, or any agent framework can call it directly, no web UI needed.
Opt any address into monthly tracking and VaultScore automatically re-checks it every month, on the anniversary of the day you started — building a real history instead of a single snapshot. Watch a wallet's discipline compound, or catch a slide before a lender does.
A REST API for your app, an MCP tool for your AI agent, and a shareable report page for your users — all backed by the exact same deterministic core, so the number never disagrees with itself depending on how you asked.
View developer docs →# REST GET /api/score/rAddress... # MCP (Claude, Cursor, any agent) score_xrpl_wallet("rAddress...") → { score, scale, reason_codes, thin_file }