What's the general consensus around ssh-agent?
It's often the case that we make much effort into securing the key but expose it via an open agent socket at a well known path or use an open control master path for multiplexing.
Certain automation and dev tools like Ansible make it a pain to work if agent key requires a confirmation.
Weirdly enough, this is _why_ I ended up using Secretive: I needed a second ssh-agent because selecting which keys are forwarded per-host is very fiddly, and while standard ssh-agent kinda supports it, 1Password doesn't. Anyway, _a_ solution for this is having separate SSH agents, and Secretive is another agent on top of 1Password.