Independent trust attestation for agentic commerce
AI agents are starting to make autonomous payments. They need to know which merchants to trust before they pay. OpenTrustToken answers that question with a signed evidence bundle, a longitudinal trust profile, and a PROCEED / CAUTION / DENY recommendation. Run a free check on any domain, or call the API before your agent transacts.
How it works
Stripe, Coinbase x402, Visa AP2, and Skyfire are all building the "how agents pay" layer. Nobody has standardized the pre-transaction question: "should this agent pay this specific merchant?" That is the layer OpenTrustToken owns.
Compositional signals, not guesses
Six signal categories run in parallel: domain age, SSL/TLS, DNS security, content analysis, reputation databases, identity verification. Each is evidence from public data. The scoring model is versioned and reproducible from stored raw data.
Signed evidence bundles, not opaque scores
Every response is cryptographically signed with Ed25519. Agents can verify the bundle without trusting us, which is why payment protocols can cite our verdicts without putting us on their critical path. The evidence bundle is the product; the score is a convenience summary over it.
Compositional brand anchor for established sites
Long-term Tranco top-50K + 5+ years of domain age + clean reputation + valid SSL is unfakeable composite evidence of established public trust. Sites meeting all four conditions receive an automatic PROCEED verdict, the same way credit bureaus weight account longevity. Any active safety signal revokes the anchor immediately.
Who this is for
Site owners and merchants
Check your trust score, see a prioritized checklist of specific improvements, and track your score history over time. Register your domain to prove ownership and unlock per-field verification points that automated checks cannot reach. Higher tiers ($29, $99, $499/mo) add business registry cross-referencing, KYC verification, and transaction insurance.
Agent developers
One API call before your agent pays. Returns a signed evidence bundle with six signal categories, a trust score, a brand tier classification (well_known / scored), a crawlability flag, an actionable checklist, and an Ed25519 signature you can verify without calling us back. Free tier: 60 requests/minute, 10,000 checks/month. Python and TypeScript SDKs available.
For developers
Verify any domain before a transaction. Signed with Ed25519 so results cannot be forged. Works today with LangChain, CrewAI, OpenAI tools, and any agent framework that can make an HTTP call.
# GET https://api.opentrusttoken.com/v1/check/merchant.com
import httpx
r = httpx.get("https://api.opentrusttoken.com/v1/check/chewy.com")
data = r.json()
# data["trustScore"] -> 77
# data["recommendation"] -> "PROCEED"
# data["brandTier"] -> "well_known"
# data["crawlability"] -> "ok" or "blocked"
# data["signals"] -> 6-category evidence breakdown
# data["flags"] -> [WELL_KNOWN_BRAND, ...]
# data["checklist"] -> actionable improvements
# data["signature"] -> Ed25519 cryptographic proof
# data["jurisdiction"] -> country + legal framework + cross-border risk
if data["recommendation"] == "DENY":
print("Refusing:", data["reasoning"])
elif data["recommendation"] == "CAUTION":
confirm_with_user(data["reasoning"])
Free tier: 60 requests/minute, 10,000/month. API docs | Integration examples | How scoring works