aws_server_control

Every piece of the infrastructure is defined in Terraform, and starting or stopping the server happens from Discord: a slash command hits a webhook, which triggers a Lambda, which handles the EC2 lifecycle. The start/stop automation is the whole point, since an instance that sits idle all week still bills like one that doesn't.

World data lives on its own EBS volume with automated snapshots, so the server survives both my own infrastructure changes and the occasional failure without losing anything.