Show HN: Sphinx – A hook that quizzes you on your AI's code before shipping

It installs via npm as sphinx-understand on Node 22.19+. Once set up, Git pauses on the branches you protect. A separate reviewer model reads the committed change through read-only git tools and never sees the conversation that wrote the code. It asks about behaviour, failure modes and invariants rather than file names, scales from one to three questions with change size. It works with Codex, Claude Code, Cursor, OpenCode, and Pi, or plain Git. The project is free and open source, and it was created by developer Parth Handa. Reaction on Hacker News leaned skeptical, with many questioning who actually benefits from being quizzed on AI-generated code, though a few see value for teams or maintainers dealing with careless AI usage. The bypass command lets a push skip the check, but every skip is logged. For founders, the notable choice is enforcing comprehension at the push step rather than assuming review happened. The README is explicit that this is practice, not a security gate: anyone with local access can skip it with sphinx yolo, which logs every skip. Whether a logged skip is enough oversight for teams shipping agent-written code is the open question.