Codecov Security Event and HashiCorp GPG Key Exposure
discuss.hashicorp.com
discuss.hashicorp.com
I understand that only signing stuff offline on a air-gapped machine is too inconvenient but I think automated (unattended) signing of release artifacts by your CI setup goes a bit too far.
While "we" as "IT folk" like to complain about things like "why is the control system of that bridge even on the Internet?" IT itself seems to be on a path where things that were once considered bad practice are now considered acceptable in the name of convenience and speed.
Either way, there is tremendous complexity in trying to achieve this. Realistically, you should take artifact signing to mean what you're installing is what the vendor actually built and not something shipped by a man in the middle. It does nothing to protect you from the vendor itself being compromised.
For questions like "why is the control system of that bridge on the Internet," consider the case a few years back when I first moved to Dallas that someone managed to compromise the city's tornado warning system because it was entirely controlled by unencrypted radio signals. The system predated the Internet. Somebody managed to set off alarms citywide for nearly an entire day by just pointing some handheld transmitter at a tower.
Unfortunately, the security posture of any infrastructure controlled by effectively anyone other than the military is close to no security at all in many cases, especially older infrastructure.
I'm curious how they'd investigate or verify this, the key in question is used for downloaded releases and plugins.