Sub-cent natural-language fact verification API for autonomous AI agents, paid via x402 USDC on Base.
What it does
AgentUtil Verify is a fact-checking API designed for autonomous AI agents. You POST a natural-language claim (up to 500 characters) and receive a structured verdict indicating whether the claim is confirmed, stale, disputed, false, or unknown, along with a confidence score (0–1), the current known truth value, data freshness, and the number of sources consulted.
The service supports five claim categories—financial, entity, geo, product, and factcheck—which can be specified explicitly or auto-detected. An urgency parameter (standard, fast, deep) controls the verification depth. Beyond the core verify endpoint, the API exposes a trending endpoint (GET /v1/trending) that lists the most-queried claims in the last 24 hours, a stats endpoint, and standard agent-discovery metadata at .well-known/agent-service.json and .well-known/agent.json.
Pricing follows the x402 protocol on Base (EIP-155:8453) using USDC. Each paid request costs $0.004 (4000 base units at 6 decimals). A free tier of 25 queries per day per IP is available with no signup required. The endpoint returned a valid x402 version 2 challenge during probing, confirming it is live.
Capabilities
Use cases
- —AI agents autonomously verifying factual claims before acting on them
- —Pipelines that cross-check LLM-generated statements against live data
- —Monitoring dashboards that flag stale or disputed financial or entity facts
- —Agent-to-agent workflows where one agent validates another's assertions
- —Automated content moderation checking user-submitted claims
Fit
Best for
- —Autonomous AI agents needing cheap, structured fact checks
- —High-volume pipelines requiring sub-cent per-query pricing
- —Developers who want zero-signup access via a free tier
- —x402-native agent ecosystems settling payments on Base
Not for
- —Deep investigative journalism requiring nuanced sourcing and editorial judgment
- —Bulk historical data retrieval or dataset construction
- —Use cases requiring guaranteed 100% accuracy with legal liability
Quick start
curl -X POST https://verify.agentutil.net/v1/verify \
-H "Content-Type: application/json" \
-d '{"claim": "The CEO of Apple is Tim Cook"}'Example
Request
{
"claim": "The capital of France is Paris",
"urgency": "standard",
"category": "geo"
}Response
{
"cached": false,
"service": "https://verify.agentutil.net",
"verdict": "confirmed",
"discover": "https://verify.agentutil.net/.well-known/agent-service.json",
"freshness": "live",
"confidence": 0.95,
"request_id": "abc123",
"source_count": 2,
"current_truth": "Paris"
}Endpoint
Quality
Full OpenAPI 3.1 schema with request/response definitions, a live x402 challenge confirming the endpoint is operational, clear pricing, and a working landing page. Docked slightly because /docs, /pricing, and /README all return 404, so there is no extended documentation beyond the OpenAPI spec and landing page.
Warnings
- —No dedicated documentation page exists (/docs returns 404)
- —No information on rate limits beyond the 25/day free tier
- —Accuracy guarantees and source provenance are not documented
Citations
- —Free tier: 25 queries/day per IP, no signup requiredhttps://verify.agentutil.net
- —Paid pricing: $0.004 per query via x402 USDC on Basehttps://verify.agentutil.net
- —x402 challenge returned with amount 4000 USDC (6 decimals) on Base (eip155:8453)https://verify.agentutil.net/v1/verify
- —Verdicts: confirmed, disputed, false, stale, unknownhttps://verify.agentutil.net
- —Categories: financial, entity, geo, product, factcheckhttps://verify.agentutil.net
- —OpenAPI spec describes POST /v1/verify, GET /v1/trending, GET /stats endpointshttps://verify.agentutil.net/openapi.json