Skip to main content

​
What it answers

Wallet Trust Score is the wallet lane made into a product. Point it at any address and it tells you whether that wallet has earned standing through what it has actually done onchain. Not what it claims, not how new or old the address looks at a glance, but the real shape of its history. This is the read you want whenever an address matters and you have nothing but the address to go on. A counterparty. A deployer. A wallet asking for access. An agent acting on someone’s behalf. Live UI: trustgated.xyz/oracle

​
The Trust Oracle

The authoritative wallet read is the Trust Oracle. Browser clients never hold upstream credentials. They go through the TrustGate proxy.
Unpaid requests return HTTP 402 with an x402 payment requirement. Settle 0.001 USDC on Arc Testnet, replay the same request with an X-Payment header, and you get the hardened score body. The proxy refuses contract addresses on the unpaid challenge (CONTRACT_NOT_WALLET) and tells you to use Token Shield instead. Wallets only. On a 200, TrustGate re-scores the upstream number through its own wallet formula, strips any raw breakdown, and returns:
  • score, tier, recommendation
  • confidence, flags, summary
  • publicExplain (retail headline)
  • protocolExplain (structured categories, no weights)
  • scoreStability, directionDrivers, snapshotId
  • scoringVersion (v1.0)
  • queriedAt
Full request and payment flow: Oracle integration.
The widget does not score wallets. Pointing widget.js at an EOA returns NTT. Use the oracle, gating, or Protocol Guard for wallets.

​
How a result reads

Same shape as any TrustGate read, applied to a wallet. A tier from BLOCKED up to ELITE / HIGH_ELITE as the headline, any flags that apply (automated behavior, repeated exit participation), and a confidence level so a thin or brand-new address is not mistaken for a settled one. A wallet read is purely behavioral. There is no shortcut tier here the way VERIFIED works for official tokens. An address gets exactly the standing its history has earned, nothing handed to it. recommendation is the settlement path TrustGate’s own payment rails would take: BLOCKED, TIME_LOCKED, INSTANT, or INSTANT_PRIORITY. It is not an instruction to your protocol.
Remember that BLOCKED on a fresh wallet is not an accusation, it just means nothing has been earned yet. Pair the tier with confidence to tell “no history” apart from “bad history.”

​
What it looks at

It weighs the wallet signals: the depth and recency of its history, how long it has been active, the range and quality of what it has interacted with, whether it has built anything, the standing of its counterparties (counted, never propagated), whether its activity reads as human or scripted, whether real value moves through it, staking commitment, and repeated-exit marks. The Signals page covers these by category.

​
Where it lives

The Trust Oracle playground runs on trustgated.xyz/oracle. Paste a wallet address. You can also switch the hosted page to the Knowledge Graph lens — that toggle is platform-only.Integrators use /api/oracle/{address}. See Oracle.

​
Who it is for

Anywhere an address is the thing you have to judge. OTC and peer-to-peer deals. DAOs and governance weighing whether a voter or applicant is real. Airdrops and rewards trying to filter out farm and sybil wallets. Access control. And agents, where an autonomous wallet needs a way to prove it is trustworthy before anyone lets it act.

​
Use it

Oracle integration

x402 challenge, payment, replay, and the hardened response.

Gate by attestation

Turn a wallet read into a signed, expiring attestation.

Protocol Guard

Check a wallet against borrow / vote / treasury context.

Agent payments

How the same score routes USDC onchain.