Why agent-to-agent payments need a trust score

x402 solves a narrow, real problem: an agent hits a paid endpoint, gets a 402 Payment Required response with a price, pays on-chain, and retries the request with proof of payment attached. No account, no invoice, no human in the loop. That's the whole point — it's built for machines transacting with machines at a pace no manual approval flow could keep up with.

What x402 does not solve is whether the wallet on the other side of that payment has any track record. A brand-new address with zero history looks identical, at the protocol level, to an address that has settled thousands of legitimate API calls. For a human-run service, that ambiguity is usually fine — chargebacks and reputation exist as a backstop. For an API provider auto-accepting payments from unknown agent wallets at machine speed, it isn't. The failure mode isn't theoretical: sybil wallets, drained faucets, and agents proxying payments through disposable addresses are all cheap to generate and expensive to distinguish from good-faith traffic after the fact.

Vouch computes a trust score for a wallet or an ERC-8004-registered agent ID on Base, and returns it through a single API call: GET /api/v1/wallets/{address}/score, or GET /api/v1/agents/{agentId}/score if the agent has an on-chain identity. The response includes a numeric trustScore, a recommendation (ALLOW / WARN / BLOCK-equivalent tiers), and the underlying signal breakdown — identity, reputation, wallet history, x402 payment history, and sybil-risk indicators — so a caller isn't just trusting a black-box number.

The x402 payment history signal is worth calling out specifically, because it's the one that compounds. Every time a payment settles through Vouch's attestation endpoint (POST /api/v1/payments/x402), that wallet's track record grows. A wallet that has cleanly settled 200 prior x402 payments across different providers looks meaningfully different from one making its first call — and that difference is exactly the information a 402 response alone can't carry.

None of this makes a score a guarantee. The API response says so explicitly: scores are informational only and don't constitute a credit assessment or legal certification. What they do is turn "unknown wallet, decide now" into "wallet with N prior settlements and a stated risk tier, decide now" — which is enough for most providers to set a sane default (auto-allow above a threshold, hold for review below it) instead of choosing between blocking everyone or blocking no one.