What music do you use to assess your system?

I don't "assess" my own system, but when I am auditioning components, I usually use these: Something off Brubeck's Take 5 record. Desmond's alto will show up a honky or forward midrange like nothing i've heard. Nora Jones Feels like Home. Music isn't the greatest, but her voice is very well recorded and up front. Will also expose midrange and vocal forwardness and lack of naturalness. James Taylor JT. Vocals reach down to the baritone range and and will expose a tubby or thin lower midrange. This is often a function of the room, but it's still a useful tool. Lambchop Is a Woman - great low-level sounds like piano pedal and fretboard sounds show a system's ability to reproduce little sounds. That's pretty much what I go by.

I try to take something that has very complex piano parts - Yelenna Eckemoff's Cold Sun comes to mind - because in my experience a system that reproduces a piano well - with all of its subtlety-to-bombast and everything in between - can do just about anything I'm going to want it to do. Very difficult instrument to reproduce truly well.

Here's mine;
Jamie Cullum -twentysomething CD/SACD (2004).
Miles Davis- Kind of Blue CD/SACD (1986, 1999).
Paul Simon -negotiations & love songs (1988).

Keep this train a rollin'

I can appreciate the fact that to evaluate a systems potential, the best recordings available would be used.
But,I actually also try to listen to the music I play and enjoy the most. Say, the best audio version recordings of CCR.

I like to listen to everything, all genres regardless of recording quality.

So I assess by randomly queuing up tracks to get a random sample of how things sounds overall.

Every recording is recorded differently, so I listen to make sure I can hear what is going on in each clearly, for better or for worse, and take note of the variations and unique aspects of each recording

Then I just listen to whatever I want to. If I hear a problem anywhere, I take note and address it, if needed.

Another benefit of using a random sample is that one is better able to determine when a particular sonic deficiency is common across multiple recordings (most likely a system problem) or not. If not, then sonic deficiencies can be attributed properly to the recording rather than the playback system