Reconfigurable intelligent surfaces (RISs) have emerged as a key enabler and promising technology for the next‐generation wireless communications by offering fine‐grained and dynamic control over electromagnetic wave propagation. However, achieving optimal and reliable performance in RIS‐assisted systems critically depends on the accurate diagnostics of faulty RIS elements. This paper proposes a novel RIS diagnostics method based on non‐uniform space‐time‐coding modulation that enables fault detection using amplitude‐only spectral measurements, eliminating the need for complex phase retrieval, coherent detection, or channel estimation. The approach involves applying a common time‐coding sequence across all RIS elements, while individually modulating each element with a unique, non‐uniform frequency. This establishes a one‐to‐one correspondence between the harmonic components in the reflected signal and the respective RIS elements. Faulty elements can be accurately identified by analyzing deviations in the measured harmonic amplitude distribution of these harmonics at their assigned modulation frequencies. Both simulation and experimental results demonstrate that the proposed method reliably detects faulty RIS elements with minimal measurement complexity, offering a computationally efficient alternative to existing diagnostic techniques.