GitHub Copilot for Pull Requests

What is GitHub Copilot for Pull Requests?

As these features are still in development, GitHub Next is actively gathering feedback and insights from the technical previews to refine and shape the future of Copilot for Pull Requests.

Copilot for Pull Requests is an AI tool developed by GitHub Next to enhance the pull request (PR) experience on GitHub. It brings the power of Copilot, an AI assistant, to help developers write better PR descriptions and facilitate smoother code review and merging processes within teams.

• Pull Request Description Suggestions: Writing a clear and comprehensive pull request description is crucial for efficient code review and collaboration. Copilot helps developers by suggesting descriptions for pull requests using OpenAI's GPT-4 model. Developers can insert specific marker tags, such as copilot:summary, copilot:walkthrough, or copilot:poem, and Copilot will generate corresponding descriptions or summaries of the changes made in the pull request.
• Test Automation Suggestions: Adding tests is essential to maintain code quality, but it can sometimes be overlooked or feel like a chore. Copilot uses AI to analyze the changes made in a PR and identifies areas that may lack tests. It then suggests tests to be built or used directly, enabling developers to catch potential issues and improve test coverage automatically.
• Ghost Text: Copilot users are familiar with "ghost text," which provides inline suggestions as they type in the code editor. GitHub is exploring bringing this UX to the pull request experience. By implementing ghost text for pull request descriptions, developers can receive helpful suggestions as they write their descriptions, ensuring they include all relevant information and follow team guidelines.

GitHub Next is continuously experimenting and exploring additional features for Copilot in the pull request workflow. Some of the ongoing projects include:

• Issue Resolution with AI: GitHub is prototyping functionality to automatically describe how to solve an issue and suggest the necessary code changes using AI. This could streamline issue resolution and simplify the process for developers.
• PR Reviews with AI: Considerable developer time is spent on code reviews and preparing PRs for review. GitHub aims to leverage AI to describe the changes in a pull request and assist with code review. AI-generated suggestions for improvement can be provided, reducing the time spent in the review cycle.
• AI-Powered PR Completion and Repair: GitHub is working on empowering its AI bot to automatically complete PRs based on minimal edits and fix common issues, such as updating unit tests or documentation, reducing the need for manual intervention.
• AI-Powered Review Responses: GitHub envisions the possibility of generating automated changes in response to specific requests during code review. AI could automatically add documentation or additional tests, optimizing the PR process.