Another way of kicking off evaluations is via the Scorecard SDK. To further automate your development workflows, you can integrate Scorecard AI by GitHub Actions.
Through this platform, evaluations are seamlessly run and integrated with your codebase, either automatically upon specific triggers or manually when needed.
DoctorGPT workflow PR trigger
When integrated, GitHub Actions automatically run evaluations for Scorecard AI each time a push or pull request is made to the repository. This is a fundamental feature, ensuring that your system under test remains consistent with expected behavior after each code change.
For instance, consider you’ve made significant changes in a feature branch and raise a pull request to merge it with the main branch. Scorecard AI, using the GitHub Actions automation, will immediately run the necessary evaluations. The results can then be viewed directly on the pull request, helping developers address any discrepancies before finalizing the merge.
To set this up:
.github/workflows/scorecard-eval.yml
file are appropriately set. The on: push:
and on: pull_request:
sections should be uncommented to enable automatic runs on these events.DoctorGPT workflow event trigger
For scenarios where you’d like to run evaluations without any code change, GitHub offers the “workflow_dispatch” event. This manual trigger allows users to start evaluations directly from the GitHub repository’s Actions tab.
Steps:
This flexibility ensures that even without a new push or PR, developers can assess system performance using Scorecard AI.
For those looking for a programmatic approach to run evaluations, Scorecard AI can trigger GitHub Actions using a repository dispatch event via the Scorecard API. This method is particularly useful when integrating evaluations as part of a more extensive CI/CD process or another system.
To utilize this:
.github/workflows/scorecard-eval.yml
file has the repository_dispatch
event configured.These diverse triggering methods support a broad range of development workflows, offering flexibility to best fit your specific needs.