Running Tests on CI

Contributors with write access to the repository can trigger full runs in the same CI systems used to produce results for The runs are triggered by pushing to branch names on the form triggers/$browser_$channel and the results will be automatically submitted to

This is useful when making infrastructure changes that could affect very many tests, in order to avoid regressions.

Note: Full runs use a lot of CI resources, so please take care to not trigger them more than necessary.


  • Base your changes on a commit for which there are already results in

  • Determine which branch name to push to by looking for refs/heads/triggers/ in .azure-pipelines.yml and .taskcluster.yml. For example, to trigger a full run of Safari Technology Preview, the branch name is triggers/safari_preview.

  • Force push to the branch, for example: git push --force-with-lease origin HEAD:triggers/safari_preview. The --force-with-lease argument is to detect if someone else has just pushed. When this happens wait for the checkout step of their triggered run to finish before you force push again.

You can see if the run started from the commit status on GitHub’s commits listing (example) and if successful the results will show up on within 10 minutes (example).