Most use of Prometheus is behind private networks, so security isn't usually a concern. The exporter endpoints are simple read-only targets, so there's not much to exploit. Also, most metrics are uninteresting from a security perspective, but there are exceptions.
Typically we recommend simple firewall rules to block public internet access, but I've also seen simple reverse proxy (nginx for example) to add TLS and some kind of auth. The Prometheus server can support SSL/x509 certs and basic auth, but I don't recommend basic auth due to the obvious clear text nature of how it works.
Adding some security (TLS + client cert auth is my obvious answer) to our standards is something we would like to do, but given the amount of people working on Prometheus, we don't have the resources to support it right now.