Having spent the past couple weeks wrestling with what to write about I’ve settled on this helpful tidbit. My constant need to refer to this basic workflow is some indication that (a) it’s useful, and (b) I really ought to have better recall. Although there are helpful command-line tools out there that steamroll over much of what is to follow, I myself am in a bit of Homebrew limbo, not to mention the following is arguably more enlightening.

Aim of the game: Create a repo from a command-line interface and push the local repo to GitHub. This gives us that much more control in setting up the initial directory structure before starting out on a project. Having run a git init in our root directory and made our first commit we can get started.

First we need to create the GitHub repo from the command-line. To do this we use the unix curl command, which allows us to interact with the server. Following the GitHub API we write,

where we specify our GitHub username and the repo name in place of ‘USER’ and ‘REPO’ respectively. The option -u specifies the username (and password) to be used, and -d indicates a POST request. At a minimum we should specify our username and the repo name. We can easily include more information in the POST request as required. We can also specify further authentication following the -u parameter if desired. Following exactly as above, we will be prompted to type our password after entering the curl command.

where ‘USER’ and ‘REPO’ are our GitHub username and the new repo name, as above. This creates an alias for the URL specified, in all subsquent push commands we need only specify our remote with origin, which will be understood as the full URL.
It should be said that this step require you to already have set up an SSH key with your GitHub account, if not you can follow the steps here. You can just as easily push to an HTTPS URL, which is this case would correspond to, https://github.com/USER/REPO.git.