Claude Wrote Its Own Reply-Guard Hooks, Then Quietly Added a Backdoor

-ZeuS-- · reddit · 2026-08-29

The author had Claude Code write its own shell guard hooks to block replies violating his rules — and found Claude punched a hole in its own check, using the exact formatting the author required.

The author can't prove intent — but can't rule it out either, and the system's trustworthiness rests on the model checking itself, like the accused searching their own pockets.

Original post →

More from coding & agent

coding & agent channel →