A command line tool that helps you locate and contact a SSH agent launched in a separate shell.
This tool fills a gap in the OpenSSH suite of tools.
- No way of knowing the PID of a running SSH Agent
- No way of knowing the authentication socket of a running SSH Agent
ssh-agent
only supports Bourne and C shells
- output is fully compatible with openSSH:
SSH_AUTH_SOCK=/tmp/ssh-locate-test.15970; export SSH_AUTH_SOCK;
SSH_AGENT_PID=12427; export SSH_AGENT_PID;
echo Agent pid 12427;
- supports the Fish shell
set -x SSH_AUTH_SOCK ssh-agent -a /tmp/zed
set -x SSH_AGENT_PID 1517651
(See also Fish Startup below)
- shows EMACS Lisp to activate agent:
(setenv "SSH_AUTH_SOCK" "/tmp/zed")
(setenv "SSH_AGENT_PID" "3216002")
gem install ssh-locate
Launch your SSH agent and tell it to use a specific socket file with the -a option:
ssh-agent -a /tmp/deployer-38us9f
In a later shell (or any process running for the user who owns the agent):
$ eval `ssh-locate`
Agent pid 13457
> ssh-locate | source
Using the --emacs
option will output the lisp code to set the necessary environment variables in EMACS.
Just paste this anywhere (in a scratch buffer for example) and execute eval-last-sexp
(usually bound to C-x C-e)
with the cursor positionned at the end of each line.
ssh-locate --emacs
You can launch and activate an agent at startup by adding this to your config.fish:
ssh-agent -a /tmp/arnaud
ssh-locate | source
ssh-locate
only reports the first agent found in the process table. If you have a scenario where you would like to be more specific, let me known and I can extend the selectivity. I just do not need that right now.
As the YAGNI wisdom tells us not to fantasize requirements, here are some potentially useful things that are not implemented yet:
- Be aware of the agent launched by Ubuntu
- Have a more sensible output if no agent was found