AI agents can now pay ChainHint per call: x402 for on-chain AML screening

TL;DR — ChainHint's wallet-reputation API now speaks x402. Beyond a free tier, an autonomous AI agent can screen any address for sanctions, mixers, entity attribution and risk across 12 chains and pay $0.10 in USDC on Base for the call — no account, no subscription. It's live today.

Subscriptions were built for humans. Agents don't sign up.

Every compliance and forensics API on the market is sold the same way: pick a tier, enter a card, pay $79 or $499 or $2,000 a month, get a key. That model assumes a human with a budget, a billing relationship, and an account.

AI agents have none of those. An agent that needs to check a counterparty wallet before sending funds doesn't want a monthly plan — it wants to pay for exactly the one check it needs, right now, and move on. The friction of "create an account, add a payment method, provision a key" is the whole reason agent commerce has been stuck.

So we removed it.

What we shipped

ChainHint's Wallet Reputation API — the endpoint that returns an address's sanctions status, mixer exposure, entity attribution and risk tier across 12 chains — now accepts payment via x402, the open HTTP-native payment protocol.

There's still a free tier (a few checks per IP per day, no key, no signup — the same one the chainhint-mcp server uses out of the box). Past that, instead of a hard "rate limit exceeded," the API answers with the standard 402 Payment Required and the terms of the check. An x402-capable agent reads those terms, signs a stablecoin payment, and retries. The check runs. One payment, one answer.

How it works

The flow is four steps and takes a couple of seconds:

1. Agent request. The agent calls the API for an address.
2. 402 Payment Required. Out of free quota, the API returns the price ($0.10), the asset (USDC), the network (Base) and where to pay.
3. Signed payment. The agent signs an EIP-3009 transferWithAuthorization — a gasless USDC transfer — and re-sends the request with an X-PAYMENT header.
4. Verify → settle → answer. The payment is verified and settled through a facilitator, and the wallet-reputation result comes back.

No gas for the agent to manage, no deposit up front, no key to rotate. The transfer is single-use at the token contract, so a payment can never be replayed for a second call. The money lands directly in our wallet on Base.

Why per-call matters for AML specifically

Counterparty screening is the textbook case for pay-per-call. An agent making an on-chain payment needs one thing at one moment: is this address safe to pay? It doesn't need a seat, a dashboard, or a contract — it needs a yes/no on a single wallet, priced like what it is: a micro-query.

Pricing it per call also makes the check honest about cost. At $0.10 a screen, running a wallet through sanctions, mixer, entity and risk checks is cheap enough to do on every counterparty, every time — which is exactly when screening actually prevents harm, rather than being reserved for the occasional manual review.

What the check actually is (and isn't)

We're a forensics tool, so the framing matters. A wallet-reputation call returns:

• Sanctions status against OFAC SDN, EU, UK, UN and Japan (MOFA) designations.
• Entity attribution from a database of 54.9M labeled addresses and 3.9M named entities.
• Category and risk tier — exchange, mixer, bridge, scam, drainer, hacker, and so on.

It's a screen against known facts, not a black-box probability. When an address isn't in our data, we say "no data," never "clean" — an absence of a listing is not a clearance. That distinction is the difference between a compliance tool and a false sense of safety.

Built for the agent internet

Making the API payable per call is one half; being discoverable is the other. ChainHint already ships as an MCP server (npx -y chainhint-mcp) so any MCP-compatible agent — Claude, Cursor, and others — can call it as a tool. With x402 on top, those same agents can now pay for it autonomously once they exhaust the free tier — and the endpoint self-registers in the x402 discovery catalog on its first paid call, so agents can find it on their own.

While most of the compliance industry still gates screening behind human subscriptions, we think the next wave of demand comes from software that pays for exactly what it uses. Agent-native screening isn't a feature bolted onto a SaaS plan — it's a different shape of product.

Try it

• Free: call the Wallet Reputation API or run npx -y chainhint-mcp — a few checks a day, no signup.
• Per call: exhaust the free tier and pay $0.10 in USDC on Base via x402.
• Human plans: still there, from $79/mo, for investigators and teams who want dashboards, tracing and full history.

The machine economy needs machine-payable infrastructure. Counterparty screening is a good place to start.

---

ChainHint is a blockchain forensics and crypto incident-response platform: paste a wallet address, get a fund-flow graph, entity attribution, sanctions screening and risk scoring in seconds. chainhint.com