Historical log of slow operations
Compass's "slowest operations" feature in the performance tab has been very helpful for debugging collection scans due to missing indexes.
The only problem is that I have to sit there and watch it to try and "catch" the slow operations with a click before they disappear. I'm running many databases on this server, and the server is talked to by many web servers running various apps and tools, and my debugging process usually starts with noticing spikes in the Operation Execution Time (from the free monitoring page).
There are obviously other ways to log slow operations, but I've found Compass so useful that I'd love to be able to do all my index debugging from within the Compass UI.
It would be useful to be able to click a "record" button, and maybe even set a threshold for the number of milliseconds required to trigger the slow operation catcher. Those operations would then get added to a list which I can come back and check later.