Introduction (2 min)

• Overview of GitHub Copilot Extensions and agents and their role in the software development life cycle
• How GitHub Copilot supports AI pair programming
• Examples of GitHub Copilot Extensions and agents in real-world development workflows

What is a GitHub Copilot Extension and Agent? (10 min)

• Explain how Copilot Extensions differ from standard Copilot suggestions
• Copilot Extensions provide context-aware, task-specific guidance rather than just inline code suggestions
• GitHub Copilot Extensions expand Copilot Chat’s functionality by enabling integration with external tools and services. These extensions can perform various tasks such as:
• Querying third-party documentation services
• Providing AI-assisted coding through third-party models
• Retrieving data from external services
• Executing specific actions in external systems
  • Importantly, GitHub Copilot Extensions are powered by GitHub Copilot Agents, which serve as the underlying extensibility platform. Copilot agents are custom tools embedded in Copilot Extensions. They integrate with Copilot Chat to provide additional functionalities tailored to specific needs.Copilot Agents provide: The ability to integrate external services and APIs into Copilot Chat
  • Support for complex integrations requiring full control over request processing
  • Flexibility to implement custom logic
  • Integration capabilities with other LLMs and the Copilot API
  • Management of conversation context
  • Complete control over user interaction handling
  • While Agents require more engineering effort and ongoing maintenance, they provide the robust foundation needed to support Copilot Extensions and enable sophisticated workflows with maximum flexibility.

How do GitHub Copilot Extensions and Agents work? (10 min)

• Components of a Copilot Agent:
• Actions: Define what tasks the agent can perform
• Directives: Instructions given to guide the agent’s behavior
• Tools: APIs or external integrations that extend functionality
• Explain how Copilot Agents leverage context (e.g., open files, repository data, user prompts)
• Describe how to create a Copilot Agent using Copilot Extensions
• Understanding API/SDK usage for extending Copilot

When to use GitHub Copilot Extensions and Agents (10 min)

GitHub Extensions use cases when you want to expand the functionality of GitHub Copilot Chat in the following way.

• Querying third-party documentation services
• Providing AI-assisted coding through third-party models
• Retrieving data from external services
• Executing specific actions in external systems

Exercise: Your first extension for GitHub Copilot (25 min)

• Guide learners through setting up a Copilot extension
• Implement a basic Copilot extension with defined actions and directives
• Showcase how to test the agent’s responses and refine its functionality

In this hands-on GitHub Skills exercise, participants will build a GitHub Copilot extension for a practical developer workflow. Learners will:

• Start a pre-configured development environment with required tools.
• Configure a web service in python or javascript to support your Copilot extension.
• Create a GitHub App and configure it to use with Copilot.
• Add custom abilities to make your extension unique.
• Try your extensions on real github.com and in VS Code.
• Optionally deploy your service to Azure and publish to the Copilot extensions marketplace.

Summary

• Recap of key points
• Sign up for 50-day challenge
• Further learning/resources
• Potential Link to Guinness attestation/badging