What this tests
parsing
Point your client at this specimen and find out — before prod does — whether it actually validates what arrives, or just hopes.
Anatomy
The body ends mid-way through a multi-byte UTF-8 character — the é that started as two bytes and arrived as one. The parser reaches the end of the stream holding half a letter, the way a translator faints mid-syllable. Not truncation at a field boundary; truncation at an ATOM boundary, which is ruder. This is what "the connection dropped" looks like at byte granularity: not a clean cut but a tear through a codepoint. Encoders that chunk without checking boundary safety produce this constantly — every naive substring(0, N) applied to a UTF-8 string produces it eventually, like a guillotine that occasionally only nicks. Clients that validate UTF-8 reject it cleanly. Clients that decode-and-replace get the U+FFFD replacement character — the little diamond of shame, �, appearing in your database where a customer's name used to end. Clients that ignore encoding entirely store the raw half-byte and pass the problem downstream, which is the traditional approach.
Replay it
curl -i https://faultydata.com/r/FD-2026-075
// response appears hereSignature
{Status:200 Headers:map[] Body: Corrupt:utf8-trunc Delay:<nil> Flaky:<nil> Sequence:<nil> Drip:<nil> StallAfter:0 Hang:false Reset:false Raw: Phases:[]}
Keep digging
FD-2026-011 — Truncated JSON body FD-2026-012 — Malformed JSON (trailing comma) FD-2026-013 — Wrong types in a valid envelope FD-2026-014 — Encoding mismatch (claims UTF-8, sends Latin-1) FD-2026-015 — UTF-8 BOM prefix FD-2026-028 — Malformed channel file update (CrowdStrike, Jul 2024)