Generating a precise & stable amplitude (and frequency) square-wave isn't difficult, see below.
https://www.eevblog.com/forum/testgear/ac-rms-dmm-tests/However extracting a precise & stable sine-wave from this square-wave may be more difficult than expected.
Since a waveform's rms value is the sum of all the individual components that make up the waveform, extracting the fundamental from a squarewave requires removing all the harmonics to a level sufficient to guarantee the "precision" of the leftover sinusoidal like waveform relative to a pure sine-wave.
Removing these harmonics entails utilizing a Low Pass filter, which must attenuate the harmonics to sufficient levels to meet the desired precision and stability of the desired resultant waveform but not attenuate the fundamental to effect the desired result.
This may become a tall task if one seeks a high level of precision, and even more so if one demands temperature and time stability. Sure one can "calibrate" the LPF and squarewave but then the circuitry must maintain an acceptable level of temp and time stability after calibration. As shown in the link above, creating a precise low frequency squarewave that is stable in frequency and amplitude over temp and time isn't difficult, converting to an acceptable precision sinusoidal waveform may be another story!!
Best,