Last released Oct 24, 2026
The exam proctor for AI coding agents: catch agents that game their tests, and measure whether their code generalizes to tests they never saw.