Trigger holdoff isn't really a good solution, and varies on some machines. Analog scopes, it's some time delay, which is fine if the modulation is constant frequency. TDS4xx's have a trigger edge count holdoff, which is nice, but it resets every time you change the scale, which is ridiculous. Modern ones I assume are actually usable.
Amplitude modulation, by definition, has no low frequency components, so we can instantly tell that HF Reject won't help at all.
We can use the trigger to demodulate the AM, to some extent. Adjust trigger level just to the highest peak. The envelope will still dance about by a few cycles (as it must*!), but you will at least get a good look at it now.
Actually detecting the AM, or starting with the modulating signal, is the only way to get a stable triggered envelope.
*In general, the carrier and envelope are not coherent (i.e., don't line up, and vary independently), so triggering on the peak will vary by a cycle of the carrier, give or take. Probably more if you can't clip the exact peak, which is very difficult even for fast, accurate triggers and a fine touch on the TRIG LVL knob. Obviously, this gets worse the more cycles there are near the peak, i.e., at large ratios of Fcarrier / Fenvelope.
The output from a synthesizer may well be coherent, in which case the display might still jitter between segments of the waveform, but only finitely many.
Tim