During the August 2024 lunar flyby of the Jupiter Icy Moons Explorer (JUICE), the MAJIS imaging spectrometer acquired the first hyperspectral observations of the Moon extending up to 5.56 µm at sub-kilometre resolution. This dataset provides an unprecedented opportunity to investigate the near-infrared thermal emission of the lunar surface and to validate MAJIS capabilities in a well-characterized planetary environment. We derive surface temperature and spectral emissivity using three independent approaches: a Bayesian inversion constrained by radiative transfer, an empirical correction based on laboratory relationships for lunar soils, and a roughness-informed thermal model that explicitly accounts for surface geometry and anisothermality. All methods reproduce the expected dependence of temperature on solar illumination, while their divergences at high incidence angles highlight the role of roughness and unresolved topography. The roughness-informed model achieves the closest agreement with thermophysical predictions, whereas the Bayesian and empirical approaches exhibit complementary strengths under different illumination regimes. Emissivity retrievals consistently reveal higher values in mare regions than in surrounding highlands, reflecting known compositional and textural contrasts, and show a wavelength-dependent inversion relative to longer-wavelength Diviner measurements. These results establish a validated framework for MAJIS thermal analysis of airless bodies and provide a benchmark for its future application to the investigation of the Jovian satellites.