An brief introduction to automation using GitHub Actions

github actions
automation
servers
workflows

The GitHub Actions feature is a powerful way of running code or a series of tasks on an independent and fresh computing environment. You can use these to automate tasks, run tasks on a schedule, test your code, or perform any number of tasks that you’d rather have a computer handle instead of doing manually.

Author

Luke W. Johnston

Published

May 21, 2024

Modified

July 9, 2024

Motivation for this micro-lesson

In our team, and in programming or software development in general, we often need to frequently run code every time a specific event happens (like someone makes changes to code) or we want to automate a task that needs to happen on a regular basis. To do that, we need to use tools that make it easier to automate those tasks. This lesson aims to share our knowledge on using one of these tools called GitHub Actions.

Assumed knowledge of reader

  • No knowledge of any specific software or programming language is needed, including Git.
  • A general understanding of what “automation” means and answering questions like:
    • How could you get a computer (conceptually) to do a task for you, instead of you doing it manually?
  • A general familiarity with GitHub, including:
    • Knowing what a Git repository, including branches, means
    • Navigating to a Git repository
    • Creating a new file (called “adding” in Git)
    • Making edits to files (“committing”)

Necessary software

  • No software is needed.
  • Need to have a GitHub account.

Learning goal

At the end of this, you will be able to describe what GitHub Actions are, identify some basic uses of for it, explain the difference between a workflow and an action, and finally create your own GitHub Actions workflow.

Take home messages

  • GitHub Actions is a feature that uses a file (called a workflow) with a list of instructions that the GitHub server uses to run to complete a set of tasks.
  • A GitHub Actions workflow is a YAML file that must be placed in the .github/workflows/ folder in a Git repository.
  • A GitHub Actions action is a reusable workflow, built by you or others, that you can find and use in your own workflows.

Lesson content

What is GitHub Actions?

At it’s core, GitHub Actions is a feature from GitHub that allows you to give a set of instructions to a server to run and complete a specific sequence of tasks. This sequence of tasks is called a workflow. Before we can continue though, we need to answer the question: What is a server?

A server is just like the computer you are using right now to read this lesson. Except that a server is a connected set of thousands of computers usually kept somewhere in a big, tightly-controlled, and managed warehouse. Servers are very powerful, large computers that is used to run complex and intensive computations, and often are used to run tasks for websites. For example, all of the Internet is found on servers across the world.

In the case of GitHub Actions, the server is GitHub’s server, and the instructions for the tasks to do are kept in a specific file (the workflow file). In order for GitHub to know about and use this file, it has to be kept in the .github/workflows/ folder of the repository and the file must be a YAML file (ending in .yml or .yaml):

reponame/
└── .github/
    └── workflows/
        ├── workflow-name-1.yml
        └── workflow-name-2.yml

In the folder setup above, there are two workflows that GitHub will run: workflow-name-1 and workflow-name-2.

An action is a custom-built workflow that you or others can develop that allows anyone to reuse a set of instructions to complete a specific task. You can think of an action as a specific type of reusable workflow. There are hundreds of pre-made actions available from GitHub, for instance, those found in their Marketplace. Because there are so many pre-made actions, you can often find one that does what you need without having to write your own, so usually your workflow would contain several actions.

General structure of a workflow

A workflow file is generally structured and organized into distinct sections:

  1. The trigger (on) section
  2. The permissions (permissions) section
  3. The jobs (jobs) section

There is also a starting name field to tell GitHub what the name or title of the action is, that is usually at the very top of the file.

The on section (required)

The on section tells GitHub when the action should run. For instance, to only run the workflow when an issue (like a bug report, a question asked, or a task to do) is created (which is called “opened”), you write:

on:
  issues:
    type: opened
Tip

The way that commands are written in official documentation often uses a shortened form. So that a . is used to separate each level. In the example above, in documentation it would be referred to as on.issues.type.

Other types of common triggers are for pull_request (when someone submits modifications to files and “requests” the repository owners to add the change to the repository) or for push (any time a change is uploaded to the repository). If you wanted to trigger on both pull requests and pushes, it would look like:

on:
  pull_request:
  push:

The permissions section (optional)

This section tells GitHub what extra permissions the workflow has. Permissions are the level of “access” a computer or person has to files. There are three permissions: none, read, and write. Read is being able to open a file, but not save it. Write is being able to open, modify, and save a file. None is no access to files at all. For security purposes, you want to limit an action’s write permission unless completely necessary, but giving read access is usually fine. This section is optional because GitHub sets the default permission to none to have the most security, but allows you to set permissions for specific things.

For instance, if you wanted to give a workflow write permission to issues only (for instance, to create a comment in an issue or to close an issue) you would write:

permissions:
  issues: write

When the permissions is set at this “top-level” (meaning it isn’t nested under another section like jobs), then the permissions given apply to all jobs within the action file. If you only wanted one job to have a specific permission, you can use permissions as a nested level item in the jobs task (more on that below).

The jobs section (required)

This section is where the actual instructions that form the core of a workflow are listed. You can write as many jobs underneath this section as you need in order to complete the task you want the GitHub servers to complete.

Each job is composed of a few subsections:

  • The name of the job in the form of job-name:.
  • The runs-on section, which tells GitHub what type of computer, more specifically the operating system, to run the job on. The mostly commonly used options are ubuntu-latest, windows-latest, or macos-latest for the three major operating systems.
  • The steps section, which is a list of instructions to run.

Within the steps section, you write a list of instructions that GitHub will run. Each instruction is a separate item in the list and is composed of a few parts:

  • The name of the step.
  • The uses section, which tells GitHub what action to use to run the step.
  • The with section, which is a list of optional inputs/parameters to give to the action from the uses section.
  • The env section, which is a list of optional environment variables to set for the step, such as the commonly used GITHUB_TOKEN variable that allows your action to interact with the GitHub API. See the note below for more information.
  • The run section, which is the actual code to run. If you use the run section, you don’t need to use the uses section. The default code that you can use in the run section is Bash code.
Tip

This GITHUB_TOKEN is a special token that GitHub provides to workflows that allows them to interact with the GitHub API. It is automatically created and available to all workflows, but you need to provide it to the workflow in order to use it. You give it to the workflow by using ${{ secrets.GITHUB_TOKEN }} in the env section. Here, the ${{ }} is a way to tell GitHub to “inject” the value of the secrets object, which also contains the GITHUB_TOKEN into the workflow. The . is what tells GitHub to look inside the secrets object for the GITHUB_TOKEN value.

An example of a job (called checkout-repository) with one step (to checkout a repository by cloning it), could look like:

jobs:
  checkout-repository:
    # Build on ubuntu, which is Linux.
    runs-on: ubuntu-latest
    steps:
      - name: Checkout the repository by cloning it.
        # A common action to use, available from GitHub.
        uses: actions/checkout@v2

You can keep chaining steps together, as many as you need, to complete the task you want your action to do.

Creating a simple workflow

Imagine you want to create a simple website, using Quarto (which we use to build this website) to build a website for you and publish it to GitHub Pages so it will be hosted online and be available to anyone. Like many tools, Quarto provides a template action file for you to use. We’ll use this as a starting point and walk through what each part of that action does.

Note

For more details and documentation on the Quarto actions and their template workflow file, you can visit their Quarto GitHub Actions repository.

What you would like is to have GitHub take the Quarto .qmd files, render them into HTML, and then publish them to GitHub Pages as a website. At least for this learning short, we will not get into any details on how to use Quarto, we’ll only add a workflow to a repository that will use Quarto to build a website.

First, we need to create a Git repository and prepare a workflow file:

  1. Create a new GitHub repository, call it anything you like. Something simple like simple-website.
  2. Clone the repository to your computer.
  3. Create a new file in the .github/workflows/ folder called build-website.yml.
  4. Open the file in a text editor and start writing out these “code” instructions.

We’ll start with the on section, as well as the name of the workflow. In this case, we want the workflow to always run whenever you or someone else pushes changes to the main branch of the repository. So we’ll start with:

name: Build website
on:
  push:
    branches:
      - main

Because we want GitHub to host the website on their GitHub Pages service, which uses a special branch called gh-pages in the repository, we need to give the workflow some permissions to be able to write to that branch. We’ll add that by using the permissions section:

permissions: 
    contents: write
    pages: write

The contents permission grants the workflow the ability to write to the repository, while the pages permission grants the workflow ability to write to the GitHub pages associated with the repository.

Next is the core part of the workflow, the jobs section. Because we want to build the contents of the repository using Quarto and then push to the gh-pages branch, we need at least three steps:

  1. Downloading the contents of the repository into the GitHub server by checking out the repository.
  2. Installing Quarto in the server.
  3. Running Quarto on the repository contents to build the website and store the output in the branch called gh-pages.

Thankfully, we don’t have to do all of these steps manually, since there already exists actions that you can download and use, such as the checkout action provided by GitHub’s actions account, along with the quarto-actions/setup and quarto-actions/publish actions provided by the Quarto team in their quarto-dev actions repository.

So we’ll first set up the jobs section with a job named build-deploy that runs-on the latest version of Ubuntu:

jobs:
  build-deploy:
    runs-on: ubuntu-latest

Next is to start the steps for the job, first of which is to check out the repository:

    steps:
      - name: Check out repository
        uses: actions/checkout@v4

Note the use of @v4 to indicate that we use the fourth version of the checkout action.

Next, we’ll install Quarto:

      - name: Set up Quarto
        uses: quarto-dev/quarto-actions/setup@v2

In this case, we need to provide the Quarto setup action (using the second version v2). Lastly, we want to run Quarto to build the website and publish it to the gh-pages branch:

      - name: Publish to GitHub Pages (and render) 
        uses: quarto-dev/quarto-actions/publish@v2
        with:
          target: gh-pages
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} 

Because we will be saving the built website to the gh-pages branch, we need to specify the output location (using the target argument accessible with the with: field) to be gh-pages. Saving to the gh-pages branch requires using the GITHUB_TOKEN environment, which is found within the secrets object.

Note

This information about the GITHUB_TOKEN is found within Quarto’s documentation as well as is included within their template file.

Your workflow should now look something like this in it’s entirety:

name: Build website
on:
  push:
    branches:
      - main
permissions: 
    contents: write
    pages: write
jobs:
  build-deploy:
    runs-on: ubuntu-latest
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Set up Quarto
        uses: quarto-dev/quarto-actions/setup@v2
      - name: Publish to GitHub Pages (and render) 
        uses: quarto-dev/quarto-actions/publish@v2
        with:
          target: gh-pages
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} 

With these few lines of code, we now have a GitHub Actions workflow that will take the Markdown files (.qmd) within the repository and build them into a website every time changes are pushed to the main branch of your repository!

Summary

  • You can use GitHub’s servers to run code or a series of tasks by through the GitHub Actions feature.
  • A workflow is a YAML file (.yml or .yaml) that is found in the .github/workflows/ folder of a Git repository.
  • One workflow file contains a list of instructions for the server to run to complete specific tasks.
  • Make use of custom actions (reusable workflows) instead of creating your own workflows from scratch.
  • Within the .github/workflows/ folder you can have many workflows doing many different things for you for the repository they are kept in.
  • There are three main sections to an workflow: on, permissions, and jobs.
    • The on section tells GitHub when the workflow should run.
    • The optional permissions section tells GitHub what extra permissions the workflow can have, if needed. The default is that the workflow will have no permissions to read or write to files in the repository.
    • The jobs section is where the actual instructions are listed that form the core of a workflow. The jobs section is composed of its own set of subsections: name, runs-on, and steps.
    • The steps sub-section of jobs is where the actual code-based instructions are put that tell GitHub what it should run. Each step is a separate item in the list and is composed of a few parts: name, uses, with, env, and run.
  • There are a large number of pre-made actions available from GitHub on their Marketplace and other sources that you can use in your own workflow.

Additional resources