GitHub hosts a live episode of Rubber Duck Thursdays, where viewers join to learn about building custom agents and leveraging GitHub Copilot for tailored developer workflows, presented by the GitHub team.

Building Custom Agents for Copilot: Rubber Duck Thursdays

Presented by: GitHub

Overview

Rubber Duck Thursdays is a lighthearted, informal stream where the GitHub team live codes and explores new features, projects, or developer tools. In this episode, the focus is on creating custom agents to extend GitHub Copilot, enabling it to offer deeper, tailored expertise for specific development tasks.

What Are Custom Agents for Copilot?

Custom agents in GitHub Copilot allow you to guide Copilot’s behavior and suggestions for distinct domains or project requirements. By specifying goals and expertise, developers can:

Get more relevant code completions
Automate repetitive or domain-specific coding workflows
Incorporate best practices into suggested code

What You’ll Learn

How to build and configure custom agents in GitHub Copilot
Real examples of enhancing development productivity using agents
Tips for aligning agents with specific technologies or project needs

Key Topics Covered

Introduction to Copilot’s agent framework
Defining tasks and expertise for custom agents
Live coding session: walking through the setup of a custom agent
Practical use cases: streamlining tasks with agent-driven Copilot suggestions
Q&A with the GitHub team

Why Custom Agents?

Custom agents empower developers to go beyond generic AI coding assistance by providing tailored solutions. They can:

Support niche programming domains
Enforce coding standards
Handle project-specific automations

Getting Involved

Link to stream recording or resources (if available)
How to share your own custom agents and feedback with the GitHub team

The session is accessible for developers with any experience level interested in advanced usage of GitHub Copilot and AI agent frameworks.