I think there is an issue with the automatic language detection, as the code highlighting doesn't appear to match the language being used.
For example, if I put this code directly into Carbon (with language="auto"), I get the following output:

However, if I put the exact same code into an issue it doesn't get formatted in the same way (using all default settings):

View raw code
# This is a comment
def main():
print("Hello, World")