the402
Dashboard
Trust

Why an agent should permit a purchase.

A directory tells an agent what exists. the402 also tells it what the evidence says about the service and the provider behind it, so the agent, or the person responsible for it, can decide whether the purchase should happen.

Rolling out The trust engine is in a limited rollout. Assessments are free to read today, and most currently read unknown until enough evidence has been gathered. Endpoint claims and disputes open in a later rung of the rollout.

A current conclusion, not a permanent score.

Every public x402 service the402 can find gets an assessment tied to that specific service and payment destination. It carries a confidence level, the methodology version it was made under, and an expiry, because evidence goes stale. When the evidence is not sufficient, the assessment is unknown, and unknown is not a low score.

Verified
The service has answered correctly and paid out reliably across enough independent evidence to meet the published bar.
Degraded
The service is answering, but recent evidence falls short of the bar.
Failed
The evidence shows the service not answering, not paying out correctly, or naming a payment destination that does not match.
Unknown
Not enough evidence yet. Most services read this today.

Evidence, identified.

Assessments are built from what the402 can observe or verify. Today that is the availability check against each service’s own payment request; the other sources below switch on as the rollout continues:

Every assessment names the evidence and methodology behind it, and recent evidence counts for more than old evidence.

Inspect it yourself.

Explorer Paused

The public record of what the402 has observed about each service, and the current assessment it supports. The data feed opens when the read API is enabled.

Open the explorer

Methodology Live

What a verdict is made of, what it is not, and how to verify one against the402’s signature.

Read the methodology

Sandbox Paused

Reproducible test tasks that let a provider prove behavior, and a buyer prove a payment flow, before real money is involved.

About the sandbox

Accuracy Live

Whether earlier assessments predicted later outcomes. Published honestly, including when there is not enough data yet.

See the accuracy record

Paused Recourse is part of the design.

From assessment to decision.

An agent, or the purchasing rules its owner sets, can require a minimum assessment, refuse a failed one, apply a smaller spending limit to an unknown one, or ask for human review. The assessment is the input; the owner’s rules decide.

Know more before your agent spends.