Use MATLAB with GitHub Actions

With GitHub® Actions, you can build and test your MATLAB® project as part of your workflow. For example, you can automatically identify any code issues in your project, run tests and generate test and coverage artifacts, and package your files into a toolbox. The GitHub actions for MATLAB let you run MATLAB code and Simulink® models on self-hosted or GitHub-hosted runners:

• To use a self-hosted runner, you must set up a computer with MATLAB as your self-hosted runner and register the runner with GitHub Actions. The runner uses the topmost MATLAB release on the system path to execute your workflow.
• To use a GitHub-hosted runner, you must include the Setup MATLAB action in your workflow to set up your preferred MATLAB release (R2021a or later) on the runner.

Overview of Actions

To run MATLAB in your workflow, use these actions when you define your workflow in the .github/workflows directory of your repository:

• To set up your GitHub Actions workflow with a specific release of MATLAB, use the Setup MATLAB action.
• To run a MATLAB build using the MATLAB build tool, use the Run MATLAB Build action.
• To run MATLAB and Simulink tests and generate artifacts, use the Run MATLAB Tests action.
• To run MATLAB scripts, functions, and statements, use the Run MATLAB Command action.

Setup MATLAB

Use the Setup MATLAB action to set up MATLAB and other MathWorks® products on a GitHub-hosted runner. The action sets up your specified MATLAB release (R2021a or later) on a Linux®, Windows®, or macOS runner. If you do not specify a release, the action sets up the latest release of MATLAB.

When you define your workflow, specify this action as matlab-actions/setup-matlab@v2. For more information, see Action for Setting Up MATLAB.

Run MATLAB Build

Use the Run MATLAB Build action to invoke the MATLAB build tool and run build tasks, such as identifying code issues, running tests, and packaging a toolbox. To use this action, you need MATLAB R2022b or a later release.

When you define your workflow, specify this action as matlab-actions/run-build@v2. For more information, see Action for Running MATLAB Builds.

Run MATLAB Tests

Use the Run MATLAB Tests action to automatically run tests authored using the MATLAB unit testing framework or Simulink Test™. You can use this action with optional inputs to generate various test and coverage artifacts.

When you define your workflow, specify this action as matlab-actions/run-tests@v2. For more information, see Action for Running MATLAB Tests.

Run MATLAB Command

Use the Run MATLAB Command action to run MATLAB scripts, functions, and statements. You can use this action to flexibly customize your test run or add a step in MATLAB to your workflow.

When you define your workflow, specify this action as matlab-actions/run-command@v2. For more information, see Action for Running MATLAB Commands.

Examples

Each example in this section provides the code that defines a workflow. To run an example, copy the example code to a workflow file and store that file in the .github/workflows directory of your repository. A workflow file can have any name, but it must have either a .yml or .yaml file extension (for example, matlab.yml).

Run Default Tasks in Build File

On a self-hosted runner, run the default tasks in a build file named buildfile.m in the root of your repository as well as all the tasks on which they depend. To run the tasks, specify the Run MATLAB Build action in your workflow. (The Run MATLAB Build action is supported in MATLAB R2022b and later.)

name: Run Default Tasks in Build File
on: [push]
jobs:
  my-job:
    name: Run MATLAB Build
    runs-on: self-hosted
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Run build
        uses: matlab-actions/run-build@v2

Generate Test and Coverage Artifacts

Using the latest release of MATLAB on a GitHub-hosted runner, run the tests in your MATLAB project and generate test results in JUnit-style XML format and code coverage results in Cobertura XML format. To set up the latest release of MATLAB on the runner, specify the Setup MATLAB action in your workflow. To run the tests and generate the artifacts, specify the Run MATLAB Tests action.

name: Generate Test and Coverage Artifacts
on: [push]
jobs:
  my-job:
    name: Run MATLAB Tests
    runs-on: ubuntu-latest
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Set up MATLAB
        uses: matlab-actions/setup-matlab@v2
      - name: Run tests
        uses: matlab-actions/run-tests@v2
        with:
          test-results-junit: test-results/results.xml
          code-coverage-cobertura: code-coverage/coverage.xml

Run MATLAB Script

Run the commands in a file named myscript.m in the root of your repository using MATLAB R2023b on a GitHub-hosted runner. To set up the specified release of MATLAB on the runner, specify the Setup MATLAB action with its release input in your workflow. To run the script, specify the Run MATLAB Command action.

name: Run MATLAB Script
on: [push]
jobs:
  my-job:
    name: Run MATLAB Script
    runs-on: ubuntu-latest
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Set up MATLAB
        uses: matlab-actions/setup-matlab@v2
        with:
          release: R2023b
      - name: Run script
        uses: matlab-actions/run-command@v2
        with:
          command: myscript

Specify MATLAB Release on Self-Hosted Runner

When you use the Run MATLAB Build, Run MATLAB Tests, or Run MATLAB Command action in your workflow, the runner uses the topmost MATLAB release on the system path. The action fails if the runner cannot find any release of MATLAB on the path.

You can prepend your preferred release of MATLAB to the PATH system environment variable of the self-hosted runner. For example, prepend MATLAB R2020b to the path and use it to run a script. The step depends on your operating system and MATLAB root folder.

name: Specify MATLAB Release on Self-Hosted Runner
on: [push]
jobs:
  my-job:
    name: Run MATLAB Script
    runs-on: self-hosted
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Prepend MATLAB to PATH on Windows (PowerShell)
        if: runner.os == 'Windows'
        run: echo "C:\Program Files\MATLAB\R2020b\bin" | Out-File -FilePath $env:GITHUB_PATH -Encoding utf8 -Append     
      - name: Prepend MATLAB to PATH on Linux
        if: runner.os == 'Linux'
        run: echo "/usr/local/MATLAB/R2020b/bin" >> $GITHUB_PATH
      - name: Prepend MATLAB to PATH on macOS
        if: runner.os == 'macOS'
        run: echo "/Applications/MATLAB_R2020b.app/bin" >> $GITHUB_PATH
      - name: Run script
        uses: matlab-actions/run-command@v2
        with:
          command: myscript

Use MATLAB Batch Licensing Token

On a GitHub-hosted runner, you need a MATLAB batch licensing token if your project is private or if your workflow includes transformation products, such as MATLAB Coder™ and MATLAB Compiler™. Batch licensing tokens are strings that enable MATLAB to start in noninteractive environments. You can request a token by submitting the MATLAB Batch Licensing Pilot form.

To use a MATLAB batch licensing token:

1. Set the token as a secret. For more information about secrets, see Using secrets in GitHub Actions.
2. Map the secret to an environment variable named MLM_LICENSE_TOKEN in your workflow.

For example, use the latest release of MATLAB on a GitHub-hosted runner to run the tests in your private project. To set up the latest release of MATLAB on the runner, specify the Setup MATLAB action in your workflow. To run the tests, specify the Run MATLAB Tests action. In this example, MyToken is the name of the secret that holds the batch licensing token.

name: Use MATLAB Batch Licensing Token
on: [push]
env:
  MLM_LICENSE_TOKEN: ${{ secrets.MyToken }}
jobs:
  my-job:
    name: Run MATLAB Tests in Private Project
    runs-on: ubuntu-latest
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Set up MATLAB
        uses: matlab-actions/setup-matlab@v2
      - name: Run tests
        uses: matlab-actions/run-tests@v2

Use Virtual Display on Linux Runner

GitHub-hosted Linux runners do not provide a display. To run MATLAB code that requires a display, such as tests that interact with an app UI, first set up a virtual display on the runner by starting an Xvfb display server. For example, set up a virtual server to run the tests for an app created in MATLAB.

name: Use Virtual Display on Linux Runner
on: [push]
jobs:
  my-job:
    name: Test an App
    runs-on: ubuntu-latest
    steps:
      - name: Check out repository
        uses: actions/checkout@v4
      - name: Start virtual display server
        if: runner.os == 'Linux'
        run: |
          sudo apt-get install -y xvfb
          Xvfb :99 &
          echo "DISPLAY=:99" >> $GITHUB_ENV
      - name: Set up MATLAB
        uses: matlab-actions/setup-matlab@v2
      - name: Run tests
        uses: matlab-actions/run-tests@v2

Notes

• To use the GitHub actions for MATLAB, enable GitHub Actions for your repository. For more information about GitHub Actions permissions, see Managing GitHub Actions settings for a repository.
• By default, when you use the Run MATLAB Build, Run MATLAB Tests, or Run MATLAB Command action, the root of your repository serves as the MATLAB startup folder. To run your MATLAB code using a different folder, specify the -sd startup option or include the cd command when using the Run MATLAB Command action.
• The Run MATLAB Build action uses the -batch option to invoke the buildtool command. In addition, in MATLAB R2019a and later, the Run MATLAB Tests and Run MATLAB Command actions use the -batch option to start MATLAB noninteractively. Preferences do not persist across different MATLAB sessions launched with the -batch option. To run code that requires the same preferences, use a single action.
• When you use the Setup MATLAB, Run MATLAB Build, Run MATLAB Tests, and Run MATLAB Command actions, you execute third-party code that is licensed under separate terms.

See Also

• Continuous Integration with MATLAB and Simulink
• Continuous Integration with MATLAB on CI Platforms

Contact Us

If you have any questions or suggestions, contact MathWorks at continuous-integration@mathworks.com.