Put a compliance firewall on your agent
No login, no API key. Paste it into a terminal and you get a real verdict: the call, a risk score, and the on-chain evidence behind each signal.
curl "https://oceanalt.com/api/risk?addr=0x0000000000000000000000000000000000000000"- Register an agent and give it a (limits / allowlist / purpose).
- Every payment runs 11 gates before settlement and gets a verdict: over-reach, hijacks, impersonation, replays and sanctioned payees are stopped before the money moves.
- The integrator acts on the verdict; direct on-chain transfers that bypass the gateway are not covered.
This gateway has 11 gates
- 1Attribution (KYA)
- 2Revocation check
- 3Identity proof (anti-impersonation)
- 4KYC gate (optional)optional
- 5Per-payment limit
- 6Daily cumulative limit
- 7Payee allowlist
- 8Mandate-intent match (anti prompt-injection)
- 9AML sanctions & risk screening
- 10Behavioral anomaly detection (optional)optional
- 11Replay & double-spend prevention
1,000 USDC
Hit a button below and watch what happens when your AI agent is attacked.
Or: call the HTTP gateway directly (zero deps, any language)
const API = "https://oceanalt.com/api/pay";
// 1) Register the agent + its own mandate (returns agentSecret)
const reg = await fetch(API, { method: "POST", headers: { "content-type": "application/json" },
body: JSON.stringify({ action: "register", agentId: "my-bot", entity: "Acme (KYC)",
mandate: { maxUsdc: 5, dailyUsdc: 50, payees: ["0x…"], purpose: "buy-data" } }) }).then(r => r.json());
// 2) Pay through the firewall (credential + one-time nonce)
const r = await fetch(API, { method: "POST",
headers: { "content-type": "application/json", "x-agent-secret": reg.agentSecret },
body: JSON.stringify({ agentId: "my-bot", amountUsdc: 2, to: "0x…", purpose: "buy-data",
nonce: crypto.randomUUID() }) }).then(r => r.json());
r.decision === "allow" ? settle() : console.warn("Blocked at", r.pillar, r.reasonEn);
