Implement AI-Driven Pull Request Reviews for Configs Repo

As the maintainer and sole developer of my configs repository, I am seeking a more efficient way to handle pull request (PR) reviews. I want to leverage AI to automate the PR review process to have a second "set of eyes" inspect my code. Because I am the only developer, right now there is no review process.. This will ensure that the quality of the code remains high, and any potential issues are identified and addressed promptly, without the need for constant manual intervention.

Technical Framework

To integrate AI-driven PR reviews, the solution should be implemented using a GitHub Actions workflow. Here are the detailed requirements:

[ ] The GitHub Actions workflow should be triggered automatically whenever a new pull request is created or an existing pull request is updated.
[ ] The AI should review the entire change set of the pull request. This includes all added, modified, and deleted lines of code to ensure a comprehensive analysis.
[ ] The AI will analyze the changes and generate a review response. This response should include:
- Identification of potential bugs or issues.
- Suggestions for improvements.
- Comments on code quality and adherence to best practices.
[ ] The review response should be posted as a comment on the pull request. The comment should clearly indicate that it is an AI-generated review for transparency.
[ ] If a new change set is pushed to the pull request, the existing AI comment should be updated to reflect the new review. If the comment cannot be updated, it should be deleted, and a new comment should be written with the updated review. This guarantees, that the comment with the AI response always relates to the latest change set.

Additional Acceptance Criteria

[ ] Document the AI review process
[ ] Perform Reviews at least for every PR that merges changes into main. Reviewing every PR (except Dependabot) would be better, but depends on cost / free tier limits.

Still to decide

[ ] Should the AI review be done in a dedicated workflow or as part of the Deployment Pipeline (after successfully linting the repo)?
[ ] Which AI tool should perform the Review? Which one is even capable? ChatGPT? Google Gemini? GitHub Copilot?

See

Desired Outcome

By implementing this AI-driven review system, I aim to achieve the following:

Efficiency: Introduce reviews by automating the process, allowing me to focus on more critical tasks. Otherwise there would be no review at all.
Consistency: Ensure that every pull request is reviewed thoroughly and consistently.
Quality: Maintain a high standard of code quality and best practices in the configs repository.
Transparency: Keep the review process transparent and understandable for contributors by clearly indicating the AI's role in the review.

By following this user story, I will be able to leverage AI to streamline the PR review process in my configs repository, ensuring efficiency, consistency, and high code quality.

sebastian-sommerfeld-io / configs