Observed evidence, paid on Base · payment rail configured

Trust data for agents that pay agents.

MCPFax joins endpoint reliability, declared payment terms, dated rating actions, and observed USDC payer graphs. It returns evidence you can dispute—not a black-box “safe” badge.

Observation archive: 11 day(s), measured from 2026-08-11 through 2026-08-21. Every window we report is bounded by that archive: nothing here is a 30- or 90-day record until the archive covers 30 or 90 days. Live figures: GET /healthcoverage.

Paid endpoints

RouteUSDC / callReturns
GET /score$0.002MCPFax credit grade, subscores, confidence, and cited evidence
GET /authenticity$0.01Payer concentration, wash-risk haircut, and authenticity-adjusted inflow for one wallet. Refuses with HTTP 422 and charges nothing when the address's own observed record does not clear the published coverage floor, rather than billing for a guaranteed not_rated.
GET /revenue$0.01Observed USDC inflow, payer breadth, trend, and cohort rank
GET /drift$0.002What changed under an integration you already wired in: payTo rotations, price and asset and network changes, terms removed, tool schemas changed. Pass up to 50 subjects and the watermark from your last call; receive the changes since it and a new watermark. Subjects MCPFax has never observed are named as unobserved rather than reported clear.
GET /payer$0.002Know-your-agent payer graph and coverage-gated screening signals
GET /payto-change-feed$0.001Observed change feed for listed x402 sellers: declared payTo changes with before/after values, other declared-terms changes (price, facilitator, fee payer), registry removals and returns, rating actions, and grade downgrades. Every response reports measured per-category coverage for the requested window.
GET /preflight$0.002Advisory preflight on one endpoint before a consequential call: a bounded live probe (reachable, status, latency, MCP initialize, current x402 price and payTo) joined to our observed history (reliability over a stated window, payTo and price changes, first seen). Returns ALLOW, WARN or UNKNOWN. Never a guarantee. No evidence, no charge.

One request

curl 'https://YOUR_HOST/score?endpoint=host:api.example.com'
# → 402 plus PAYMENT-REQUIRED (x402 v1, network "base")
# Retry with X-PAYMENT, or the equivalent alias PAYMENT-SIGNATURE.

Free surfaces

GET /health readiness and coverage · GET /llms.txt agent-readable docs · POST /mcp MCP initialize and tool discovery. Tool calls use the same x402 prices. Licensees use GET /feed/reliability with a bearer key and cursor pagination.

What the numbers do not prove. A wallet balance is not revenue. Incoming USDC is observed on-chain but is not yet guaranteed to be an x402 sale. HHI measures concentration, not common control. Funding-source edges are still missing, so observed organic/wash percentages remain null. Wash-risk and payer-concentration verdicts are withheld as “insufficient payer observations — not rated” until a subject’s own observed payment record covers at least 14 days and 20 payments—so a thin record is reported as thin, never as a finding against a named party.

MCPFax Credit Model 1.0 is ordinal, not a calibrated probability of default, and it is not validated over any forward horizon. Observation archive: 11 day(s), measured from 2026-08-11 through 2026-08-21. See the response’s evidence citations and coverage block before acting.