Open metrics · updated 2026-07-27

Real numbers, no projections

Everything on this page is verifiable against the linked source. If a number isn't measurable yet, it says so instead of estimating.

Engine

0.5.0
x402-conformance-suite on PyPI (Apache-2.0)
203
tests in the engine suite, all passing at release
7
checks in a standard audit (4 before 0.5.0)
27
production endpoints in the v0.3 validation corpus, audited in 5.2 s

The corpus result that matters

In the 2026-07-27 corpus run, 0 of 27 endpoints passed all four v0.3 checks. That is not a typo — most x402 endpoints in the wild are not conformant. Two of the 27 failed only CAIP-2 at the time; both pass today after the 0.4.1 parser fix below. Full matrix: VALIDATION_REPORT_v0.3.md.

The bug we shipped (and how it went)

Our pre-0.4.1 CAIP-2 parser read the network only at the top level of the payment payload. x402 v2 carries it at accepts[].network — so every strict-v2 endpoint false-failed. Worse, we had already emailed merchants audit results containing that false FAIL.

A merchant replied with a precise diagnosis (free catalog root, paid products, network in accepts[]). Same day: reproduced live, fixed the parser, added manifest/product network fallbacks, published 0.4.1, redeployed, re-audited, and sent correction emails to the two merchants whose results changed. The full diff and tests are in the public repo history.

Business

5
pro keys issued — all free 3-month early-adopter grants
$0
paying-customer revenue to date
3/day
free audits per IP; Pro is $9/mo, Enterprise $49/mo
1
person running this (Gael L Chulim), hosted on Render, billed via Stripe

What we don't track

The API keeps no audit counter, no per-user history, and stores no audit results — only rate-limit timestamps per IP (in-memory, gone on restart) and issued API keys. When we can measure audits-served honestly, that number will appear here. Until then, it won't.

Sources