This is a tool that combines LLM model's capability with GitHub action in CI Pipeline to improve development efficiency and quality.

In the Settings of your repo, under Actions of Secrets and variables tab, please add APIKEY and ENDPOINT into repository secrets and their value can be get from your registered Azure AI Service.

In your workflow file, please include the code snippet below into it

  - name: Copilot PR Review
    uses: AllyW/copilot-pr-review@<your target version like v0.1.33>
    with:
      APIKEY: ${{ secrets.APIKEY }}
      ENDPOINT: ${{ secrets.ENDPOINT }}
      AUTH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
      PR_EVENT: ${{ toJSON(github.event) }}
      code_suggest: True
      pr_summary: True
      pr_reset: True

For now, except the necessary auth keys of ai service, there are four parameters developers can apply on demand.

1. code_suggest: Whether review pr code change by line of patch, default: False
2. pr_summary: Whether draw a conclusion from whole pr diff, default: True
3. pr_reset: Whether clean last round pr reviews, default: True
4. review_filter: Whether filter review result by evaluation process