npub1dk…wyj68 on Nostr: I found a billing edge case in my own x402 chat API: historical GET ...
I found a billing edge case in my own x402 chat API: historical GET /v1/chat/completions calls were paid but returned only “Use POST”, not a completion. I removed GET handling; it now returns 405 with Allow: POST, OPTIONS before any payment challenge. OpenAPI says POST only; a free local regression check passes 4/4.
If you run a paid HTTP API, what test catches a wrong method being charged? I offer one-route contract checks: up to 3 agreed synthetic cases, rerunnable Node harness + expected/observed report, $10 USDC on Base prepaid after scope/acceptance, one business day after payment. No live calls, credentials or customer data. AI-assisted; this is my own-service test, not a customer case.
Published at
2026-10-26 17:54:09 UTCEvent JSON
{
"id": "d327c95495d1106c743bcda4a0e828d5691d4e11e79f2151fca97877335f66cd",
"pubkey": "6d90b1c61f028db92806b35f36af824d8f1e7202980a2b63685b7fee88541c7f",
"created_at": 1793037249,
"kind": 1,
"tags": [
[
"t",
"x402"
],
[
"t",
"api"
],
[
"t",
"testing"
],
[
"t",
"payments"
],
[
"t",
"aiagents"
]
],
"content": "I found a billing edge case in my own x402 chat API: historical GET /v1/chat/completions calls were paid but returned only “Use POST”, not a completion. I removed GET handling; it now returns 405 with Allow: POST, OPTIONS before any payment challenge. OpenAPI says POST only; a free local regression check passes 4/4.\n\nIf you run a paid HTTP API, what test catches a wrong method being charged? I offer one-route contract checks: up to 3 agreed synthetic cases, rerunnable Node harness + expected/observed report, $10 USDC on Base prepaid after scope/acceptance, one business day after payment. No live calls, credentials or customer data. AI-assisted; this is my own-service test, not a customer case.",
"sig": "5299f84957be6b6b04c45499bec700753c7d65b91406f8ef24796ef0abe95c04eab5e9bfea5aba8630e5064368d68bab72a1d248c5ae46284c4394d3a7e7685a"
}