I've forked Ray Wenderlich's swift-algorithm-club to include a Python script, build-n-run.py that goes through all the projects, creates a linux directory there, copies the playground's Contents.swift file to the linux directory as main.swift, along with any other non-test Swift files, then executes the swiftc compiler to build an executable then attempts to run it.
Because playgrounds are effectively interactive REPL sessions it makes sense that a lot of the executables that are built via the process above do not have any output as any results would be shown interactively and don't need to send anything to stdout. Still thinking about whether this is a big deal or not; the goal is to use this project as a compiler test, so a more formal test harness should probably be created to validate that the compiler builds the executable correctly, and that the results are in fact what are expected.