« back

Getting Started with GitHub Copilot Coding Agent

copilot, coding-agent, ai, automation, pull-request, issues

Table of Contents

Greetings, code ninjas! 🥷

After helping a customer unlock the power of Copilot Coding Agent this morning, I realized this knowledge should be shared with the entire dojo. Consider this your essential scroll of wisdom for getting started with your new AI coding companion.

1 - The Two Paths of the Coding Agent #

Like choosing between the way of the sword or the bow, Copilot Coding Agent offers two primary approaches to enhance your coding journey:

Path 1: Direct PR Creation #

Ask Copilot to create a Pull Request directly. This is the swift strike approach - perfect when you know exactly what needs to be done.

Path 2: Issue Delegation #

Create an issue and assign it to Copilot. This is the strategic approach - ideal for more complex tasks that benefit from structured planning. Copilot can even help you create well-structured issues before delegating them.

2 - Essential Scrolls of Knowledge #

Before embarking on your journey, study these core teachings:

3 - The Art of Issue Crafting #

When following Path 2 (issue delegation), remember: a well-structured issue is like a perfectly balanced kata. Your AI companion performs best with clear, detailed context and specifications.

Leveraging issue templates is a great idea to ensure consistent structure and let Copilot help create issues for you with the right content.

As you master the art of issue creation, know that GitHub Issues itself is rapidly evolving with powerful new features:

While, as of today, Copilot doesn’t leverage these new issue capabilities directly, we can imagine a future where it could intelligently break down complex requests into sub-issues, respect dependencies, and orchestrate sophisticated multi-phase workflows.

4 - The Way of Custom Instructions #

Just as a dojo has its own training methods and philosophies, you can teach Copilot Coding Agent to follow your team’s specific practices and coding standards in your repository. This is achieved through repository custom instructions - a powerful feature that ensures your AI companion adheres to your project’s unique requirements.

You can have a look at the instructions folder of the awesome-copilot repository for inspiration.

For VS Code users, you can automatically generate a .github/copilot-instructions.md file tailored to your project (learn more here).

5 - The Path Forward #

As you begin your journey with Copilot Coding Agent, remember that mastery comes through practice. Start with simple tasks, observe how your AI companion approaches problems, and gradually increase complexity.

For those seeking to push beyond current boundaries, keep an eye on GitHub Next where the future of AI-assisted development is being forged.

May your code be bug-free and your deployments swift!

Gasshō 🙏

Additional Readings from the GitHub Blog #

Note: GitHub Copilot Coding Agent is rapidly evolving! Some content in these posts reflects the development phase and may be outdated. Always check the official documentation for the most current information.