Still hard to imagine
Not that I'm disagreeing with you really - mechanical devices have high failure rates. 10 years maintenance-free is a bit of a dream. As a mechanical engineer I'm interested to see whether the Stirling engines involved have radically different scaling to optimise for reliability, or NASA just plan to do the engineering really well.
Of course manned maintenance may be possible - they are touting this technology for Mars bases. Plus robotic maintenance is going to become more of a thing over time.
Note that while implementations usually show a crank and rod with a flywheel, it could just as easily use a magnet and coil to generate electricity.
That get's you down to a single part.
Then you have this:
https://en.wikipedia.org/wiki/Thermoacoustic_heat_engine
...that gets you down to something that can generate sound from a heat differential, and you could couple that sound to some kind of transducer to generate electricity. Still a moving part, though.
You probably can't get zero moving parts and yet have it do useful work, but you can get really close I think.
As a heuristic, the less mechanical parts in any system, the more efficient it is over time.
Ideally, everything that's built is engineered like a spacecraft, because to some extent it is, and is onboard one.