What this tests
semanticsparsing
Point your client at this specimen and find out — before prod does — whether it interprets status codes and content types correctly, or makes things up.
Anatomy
You sent amount=100, currency=USD. The response is a cheerful 200 confirming amount=10000, currency=USD — order of magnitude off, field order shuffled, a value that resembles yours the way a ransom note resembles a grocery list. The API heard you. It heard you wrong, and it confirmed confidently. Knight Capital lost $440 million in 45 minutes to a deploy that routed orders to dead test code — not an error response, a wrong response, confirmed successfully at machine speed. The most expensive class of bug in existence: not "it failed" but "it succeeded incorrectly", which only gets discovered by the market, the auditors, or the front page. Clients that log the request but not the echoed confirmation will reconcile at month-end and find a discrepancy that no error log explains — because nothing errored. The specimen confirms the wrong amount. Whether your client notices the echo is the entire exam.
Replay it
curl -i https://faultydata.com/r/FD-2026-072
// response appears hereSignature
{Status:200 Headers:map[] Body:{"confirmed":true,"order_id":"ORD-88231","amount":10000,"currency":"USD","note":"you sent 100"} Corrupt: Delay:<nil> Flaky:<nil> Sequence:<nil> Drip:<nil> StallAfter:0 Hang:false Reset:false Raw: Phases:[]}
Keep digging
FD-2026-013 — Wrong types in a valid envelope FD-2026-034 — HTTP 200 wrapping an error (GraphQL/SOAP pattern) FD-2026-035 — Stack trace in the error body FD-2026-053 — SSO login page instead of your API response FD-2026-062 — Migration returns other customers' data (TSB, Apr 2018) FD-2026-087 — 200 OK, body never arrived