Updating ssh-agent state from within Emacs

Question

I have recently switched from launching Emacs from the terminal (i.e. emacs --init-directory "path/to/my/init") to launching it from a universal keybinding (i.e. super-e). This has had the unintended side-effect of making Magit essentially unusable, because it is no longer 'inheriting' the state of my ssh-agent. It used to be that if I was getting Permission denied (publickey) errors, it was a matter of a closing Emacs, ensuring my agent was running and the right keys were added to it, then relaunching Emacs from that terminal session.

However, when I launch it from the keybinding, I'm not able to get it to recognize ssh-agent state. That it can't pick this up from a terminal session that had nothing to do with its launch isn't too surprising, but I thought that running ssh-agent from within a term window might address the problem. No such luck.

I suppose that daemonizing the ssh-agent setup would get around this problem, but I'm unsure if I want to go that route. What options are available to me for configuring this within Emacs?

Answer 1

Well you've identified the issue: You're not starting Emacs in an environment which knows about your ssh-agent, so the most obvious solution is to remedy that.

Either start an agent as part of your desktop session, or else use a command or wrapper script which exports those environment variables for starting Emacs.

Keychain can help:

• https://www.funtoo.org/Keychain

You can also just C-hf setenv the environment variables in Emacs. If your shell will have the environment, then you can tell Emacs to run a shell command to obtain the values. If you go that route, you might like https://github.com/purcell/exec-path-from-shell