How It Works

Trunk Check integrates with GitHub to automatically identify linter issues, unformatted files, and vulnerabilities in your repositories without ever sending your code to Trunk.

If you don't use GitHub, we recommend you check out the Continuous Integration guide.

How it works

Trunk Check's GitHub integrations rely on the following:

  • An installation of the GitHub app in your GitHub organization, and

  • A .trunk repository in your GitHub organization.

What is a .trunk repository?

The .trunk repository contains the workflows run to scan your codebase and pull requests. We recommend creating a .trunk repository in your GitHub organization using this template repository.

Your .trunk repository must be added to your Trunk GitHub app installation. You can verify this by navigating to:<your_organization>/settings/installations, clicking "configure" next to Trunk-io, and verifying that the repository access is either "All repositories" or that your .trunk repository is selected.

To find Check issues in your repositories and pull requests, we dispatch GitHub Actions workflows in your .trunk repository, which check out your repositories and pull requests and then run trunk check in them. This strategy allows you to:

  • start using Trunk Check in all your repositories without any configuration, and

  • be in full control over the environment where we analyze your code, since we're running on your GitHub Actions runners.

🚧 .trunk should have private visibility

Since we use workflow runs in .trunk to analyze any repository in your organization and record Check findings, you should think carefully about who has permissions to view workflow runs in your .trunk repository. For most organizations, simply making your .trunk repository private will be sufficient.

If you want to version the linter configuration for a given repo or enable linters that require more manual configuration, you can always create and commit your Trunk configuration in said repository.

Checking pull requests

Trunk Check can automatically detect new Check issues on your pull requests and flag them so that you can prevent pull requests from introducing any new issues in your repository.

When running on a pull request, Trunk Check will only flag new issues, not existing ones, so that your engineers don't have to fix pre-existing linter issues in every file they touch - this is the same hold-the-line technology that our VSCode extension and CLI use.

Fixing issues in pull requests

To confirm that you've fixed issues identified by Trunk Check before pushing your pull request, just run trunk check.

If Trunk continues to identify new Check issues on your PR, first try merging the latest changes from your base branch. When Trunk runs on a PR, it runs on a commit that merges your PR into its base branch, just like GitHub workflows.

If this continues to fail, then run git checkout refs/pull/<PR number>/merge && trunk check. This is a reference to the merge commit GitHub creates.

Skipping Trunk Check

You can include /trunk skip-check in the body of a PR description (i.e. the first comment on a given PR) to mark Trunk Check as "skipped". Trunk Check will still run on your PR and report issues, but this will allow the PR to pass a GitHub required status check on Trunk Check.

This can be helpful if Check is flagging known issues in a given PR which you don't want to ignore, which if you're doing a large refactor, can come in very handy.

If you don't want Trunk Check to run on pull requests, turn it off in your repository's settings.

Scanning your repository

Trunk Check can scan your repository for Check issues on a daily cadence, upload them to Trunk for you to review at your convenience, and notify you via Slack whenever new issues are discovered in your repository.

This allows you to build confidence in the code health of your repositories:

  • You will be alerted quickly in a Heartbleed-type event, giving you assurances about whether or not a newly discovered vulnerability affects any of your repositories, and

  • You can monitor how many Check issues exist in each of your repositories and make data-driven decisions about prioritizing efforts to reduce tech debt

If you don't want Trunk Check to scan your repository on a daily cadence or notify you, you can turn it off in your repository's settings.

(optional) Custom setup logic

If you need to do some setup before trunk check runs in your-org/your-repo, you can define a GitHub composite action in .trunk/setup-ci/action.yaml in your-repo. This can be important if, for example, a linter needs some generated code to be present before it can run:

name: Trunk Check setup
description: Set up dependencies for Trunk Check

  using: composite
    - name: Build required trunk check inputs
      shell: bash
      run: bazel build ... --build_tag_filters=pre-lint
    - name: Install eslint dependencies
      shell: bash
      run: npm install

Read more in the documentation for our GitHub Action.

Last updated