Skip to content

The trust layer for
x402 agent payments

Inspect x402 endpoint challenges, payment terms and recorded receiver signals before your agent decides to pay. A valid challenge is an observation, not a guarantee of current availability or delivery of the purchased result.

65% of 7,546 endpoint URLs listed as healthy by 402index returned a valid x402 challenge in their latest observations (4,913 confirmed). This is an endpoint comparison, not a count of unique services.
3920unique domains with valid latest observations
46646payable endpoints
31%dead/invalid among probed endpoints

Observation age · Median: 72 h · Oldest live observation: 22 days.

Latest crawl activity: . Age summary computed: .

Live describes the moment an endpoint was observed. The registry is checked in rotation; lastCrawl is not the last check of every endpoint. Inspect the per-record lastChecked before relying on a cached verdict.

31% dead or invalid among 69,719 probed endpoints. probed endpoints only; unprobedCandidates, blockedByPolicy and methodMismatch are excluded, not counted as dead

What you get

Trust & safety oracle $0.004

Before paying an unknown x402 endpoint: liveness, uptime/reputation, a scam & anomaly scan (payTo hijack, price bait-and-switch, honeypot, non-USDC asset), on-chain receiver profile, and a pay/avoid verdict.

GET /trust?endpoint=… · free /verify

Payment guard SDK npm

Checks PulseFeed's cached verdict before calling your paying fetch. By default, unknown endpoints and verification failures are allowed. Set explicit block policies when a missing verdict must stop the request. Zero dependencies.

npm i pulsefeed-x402-guard

MCP server npm

Give Claude Desktop / Cursor / Cline tools to find live x402 services and verify endpoints before paying.

npx -y pulsefeed-x402-mcp

Agent-framework tools npm

Drop-in tools for the Vercel AI SDK and LangChain: verify an x402 endpoint before your agent pays, and discover working services.

npm i pulsefeed-x402-ai-tools

On-chain intelligence $0.01+

Real-time Base signals for agents: token pulse, whale alerts, smart-money accumulation/distribution, momentum.

/pulse · /whales · /flow · /momentum

For AI agents — verify before you pay

// Stop on unknown endpoints or a failed cached check.
import { guardFetch } from "pulsefeed-x402-guard";
const guarded = guardFetch(payingFetch, {
  onUnknown: "block",
  onError: "block",
});
await guarded(endpoint);

This example blocks avoid, unknown endpoints and verification failures. Known caution verdicts are allowed unless you add them to block. SDK defaults allow unknown endpoints and verification failures. Cached results describe the last observation; they do not guarantee current availability or delivery of a purchased result.

Or one call: GET https://pulsefeed.dev/verify?endpoint=<url> (free) · GET https://pulsefeed.dev/trust?endpoint=<url> (live, paid).

Pricing (pay-per-call, x402 · USDC on Base)

EndpointWhatPrice
/verifyCached trust verdictfree
/trustLive verification + scam scan + on-chain$0.004
/trust/historyEndpoint track record over time$0.008
/trust/datasetBulk trust dataset (all services, one call)$0.10
/mcp/drift/historyFull MCP supply-chain drift series since 2026-07-30$0.10
/data/fullCross-domain dataset (x402 + MCP + incidents)$1.00
/pulse · /momentumToken activity / momentum$0.01
/whales · /flowWhale alerts / smart-money flow$0.02

PulseFeed — independent trust & safety layer for the x402 agent economy.
Research: frozen dataset with recount scripts — doi:10.5281/zenodo.22251005 (concept DOI, resolves to the latest version; CC BY 4.0) · daily manifests signed (Ed25519), timestamped via OpenTimestamps and upgraded to a Bitcoin block proof once a block confirms (the newest day is a calendar receipt until then) — /anchors/latest.json · verdict freshness published in /status.json.
Contact: contact page · nikolife2016@gmail.com · GitHub · security.txt
Observatory · Changes · State of x402 · Trust Score · Embed widget · Methodology · OpenAPI · llms.txt · manifest · guard SDK · MCP · AI SDK tools