Neat. But why reinvent a bad version of matplotlib when all you need to make is an MPL backend? Is it very hard to make a backend?
I actually use youplot[1] for this very reason so I can pipe small datasets into an exploratory visualization straight from the command line.