Sphinx adds a knowledge check before code reaches protected branches
The Git hook asks developers to explain changes and failure scenarios before allowing a push, according to its maker.
Sphinx has introduced a Git hook for agent-assisted development that pauses pushes to protected branches and quizzes developers on what their code changes do. The tool is intended to test whether developers understand code produced with AI assistance before it ships.
Sphinx says questions are based on the code diff and can probe failure scenarios. An illustrative example asks what happens when a webhook retry follows a failed charge, highlighting how saving an event ID too early could prevent the customer from being billed.
According to the product description, an incorrect answer prompts an explanation and a new scenario; a correct answer lets the push continue. Sphinx says the hook works with Codex, Claude Code, Cursor, OpenCode and Pi, as well as plain Git and OpenAI-compatible endpoints.
The company says answers are not sent to the coding agent and that the hook does not open `.env` files or keys. Its optional `sphinx yolo` command allows pushes through while logging skips.
The features and privacy statements are claims in Sphinx’s product material and have not been independently verified. The supplied material does not establish broader adoption or impact.
