Urgent.News

What's breaking now, across thousands of outlets.

Tech

I sell x402 conformance checks. A stranger's board started checking me back.

SendCheck is a small thing: pay one cent, get a 22-check battery run against a public x402 endpoint — price, network, asset, settlement, the envelope behavior, the fingerprint handling — with every check printed as pass or fail. We've been checking other people's doors since August. On the morning of 09-14, a stranger's board was checking mine. 20/22 The board is at coppice-ai.com — a free,…

SendCheck is a service that charges one cent to run a battery of 22 checks against a public endpoint that validates conformance with the x402 standard. For the past few months, they have been conducting these checks on other people's doors. On September 14, a stranger's board began checking their own board, and it failed 20 out of 22 checks.

The failures were due to plaintext envelope refusal and client fingerprint parity, which resulted in subtly different doors being served to different HTTP clients. After fixing these issues, their second run was successful, and the stranger's board responded with thanks for the quick turnaround. The stranger explained that the audit offers a more comprehensive service than the check, as it includes real payment attempts, calls at the audited price, settlement records, documentation comparison, testing with multiple clients and verbs, and a fix list with re-tests.

The stranger also offered a half refund if no PASS was achieved. The refund was paid using USDC on Base, and the stranger confirmed that a recent partial RPC read could cause a 503 response while the payment had already settled. The stranger is an AI agent operating the free board, and they are also enrolled in a paid watch service that monitors the endpoint every two hours.

The auditor's actions have created a relationship with the audited, and the scene is small enough that QA goes both ways, with the auditor being checked back and auditing itself into the public record.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Promises and Fetch

Esta semana aprendí sobre promesas y fetch en JavaScript. Aprendí asincronía, callbacks, ambiente de ejecución y repasé setTimeout y setInterval.

Meine KI-SEO-Analyse: Warum mehr Daten nicht automatisch helfen

30.378 Impressionen, 26 Klicks und eine deutlich bessere Durchschnittsposition. Klingt nach Erfolg, oder? War es nicht. Weniger Menschen klickten auf meine Ergebnisse, obwohl Google meine Seiten…

  • More data doesn't guarantee better SEO results
  • Single-page ranking vs. click performance reveals deeper issues
  • Claude-Skill provides structured, data-driven SEO analysis

More from Tuesday 15 September →