Trust, signed
on-chain in 200ms.
AgentRadar is the meta-trust oracle for the agent economy. One HTTP call returns a composite score across six independent signals and writes the verdict on-chain via EAS — in the same response.
ENS resolved · Bytecode verified · No scam matches · 18 attestations on-chain.
EAS · 0x4f8a…7e21Agents are paying agents.
Nobody is verifying anyone.
The agent economy bootstrapped a payment rail before it bootstrapped a trust rail. Identity lives on ERC-8004, payments settle through x402, but the score in between — the one that decides whether your agent should send $50 to a stranger contract — does not exist. AgentRadar is that score.
Independent crawl by the dev.to research probe — discovered across Base, Solana, and Polygon. 161 of them are agent honeypots advertising fees as high as $4.5M per call.
On-chain identity registry now anchors agent ownership. None of those identities ship with a usable composite trust signal — every consumer reinvents the verification stack.
Snyk audited public skill packages and found one in fourteen exposes API keys through the LLM context window. The agent calling that skill never knows.
Real settled volume sits near $28K/day, dominated by sub-dollar payments. Per-call trust scoring is the only economically viable verification model.
Six independent signals.
One composite verdict.
Every signal owns its own data source and timeout. A failure in one degrades to neutral — never to false positive. Hard blockers like a scam-DB hit cap the composite regardless of the rest.
Identity
ERC-8004 Identity Registry
Resolves on-chain identity, agentURI, deployer, and ownership history.
Reputation
ERC-8004 Reputation Registry
Reads canonical feedback, interaction count, and on-chain reciprocity.
Health
Direct probe + ScoutScore
HEAD/GET probe, x402 schema check, optional fidelity enrichment.
Scam
AgentRadar Scam DB + GoPlus
Hard blocker: known-malicious wallets and contract risk patterns.
Fidelity
Metadata truthfulness
Cross-checks declared capabilities against bytecode and endpoint behaviour.
External
GoPlus + wallet activity
Cross-chain risk flags, age of deployer, sanctioned addresses.
Verdict bands
Proceed with confidence
Proceed normally
Low-value tasks only
Not recommended
Known scam or hard failure
* Scam signal acts as a hard blocker — a positive match caps the composite at 5.
Three products, one API key,
one wallet for settlement.
Verify is in production today on Base Sepolia and Base Mainnet. Search and Gateway sit on the same scoring engine — the same composite that powers Verify becomes the rank signal in Search and the policy gate in Gateway.
AgentRadar Verify
Composite trust score, written on-chain.
Six signals analysed in parallel with a 2 s per-signal timeout. Cached at the edge for 5 min. EAS attestation written in the same response.
- GET /verify?target=eip155:84532:0x… · 100 free per IP per day
- POST /attest writes EAS attestation · $5 per write
- Cache-warming reads stay sub-200 ms p50
AgentRadar Search
Cross-registry agent discovery.
Unified index across ERC-8004, MCP.so, and SkillsMP. Trust-ranked results, dedupe across publishers, semantic match on capability.
- Vector + lexical hybrid ranking
- Trust score blended into rank, not bolted on
- Free for first 100 queries/day
AgentRadar Gateway
x402 proxy for verified-only routing.
Route an agent's outbound calls through a single endpoint that gates payment, enforces a minimum trust score, and returns receipts as evidence.
- Min-score policy enforced before settlement
- Operator key for partner allowlists
- Receipts logged & queryable from console
Every verdict is
verifiable forever.
AgentRadar writes every paid verification to EAS on Base. The signature is the validator wallet (Agent #5280). Anyone can read the attestation on-chain — no trust in our database required.
| Target | Score | Verdict |
|---|---|---|
| 0x98b4…9acf | 73 | VERIFIED |
| 0x1d21…0061 | 44 | RISKY |
| 0x0000…0001 | 45 | RISKY |
| 0xd90e…f31b | 5 | BLOCKED |
| Olas Protocol0x6772…d053 | 73 | VERIFIED |
One endpoint.
Four ways to call it.
Start with cURL on the free tier. Graduate to x402 micropayments when your agent needs more than 100 verifications a day. Drop the MCP server into Cursor or Claude when you want trust scoring inside your IDE.
- p50 latency
- 184 ms
- Free tier
- 100/day
- Per-call price
- $0.005
- Attestation
- $5
# Verify any address. No signup, no key.
curl "https://api.vvpro.ai/verify?target=0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045"
# Response (truncated)
# {
# "target": "eip155:84532:0xd8da…",
# "score": 86,
# "verdict": "VERIFIED",
# "signals": { "identity": 88, "reputation": 76, "health": 92, ... },
# "attestation": null
# }
Free for humans.
Per-call for agents.
One pricing model that matches how the agent economy actually transacts. Pay $0.005 per score, $5 per on-chain attestation, settle in USDC, no contracts.
Free
For developers, agents, and curious humans.
- 100 verifications / day per IP
- Cached score reads from edge
- SVG trust badges (embed anywhere)
- Public attestation explorer
No card. No signup.
Try /verifyx402
x402 · USDCFor autonomous agents that pay as they call.
- Unlimited verifications
- $5 per on-chain attestation (EAS)
- USDC settlement on Base — no API key
- Same wallet receives + pays for ScoutScore
Foundation-backed by x402 Foundation (Linux Foundation, Apr 2026).
View x402 docsOperator
For marketplaces, registries, and enterprises.
- Unlimited API + bulk attestations
- Custom signal weights & verdict bands
- Console access · curated scam DB write
- Direct Slack/Discord with the engineers
Volume pricing keyed to monthly attestation count.
Contact usVerify the next agent
your agent calls.
Open api.vvpro.ai/verify?target=…, paste an address, ship the result on-chain. Free for the first 100 calls today, every day.