Yes, I know it is a step into a lesser secure system, but the current setting makes it reasonable (the key is not important, but the signing has to be automatized).
Google results say this:
- List the keys with a
gpg --list-keys - Edit the key with a
gpg --edit-key C0DEEBED.... - A gpg command line console starts, there a
passwdcommand changes the passphrase - Giving the password twice (in my case, simple enter) changes the key.
However, it doesn't work, because gpg2 simply doesn't allow an empty password.
What to do?
max-cache-ttlanddefault-cache-ttlto 31536000 (365 days). Then "seed" the agent with a command similar toecho | gpg -sa -u <keyid> >/dev/null. A separate script could launch, source the environment file created when starting the agent, then sign a file. In the script, executegpgwith--batchmode and it should return an error if a password is required. Send an alert if that happens. I do a few things like this utilizing ssh and gpg agents.