GHU: Building AI Agents with VS Code and GitHub

From the creator

In this interactive session, you’ll build and deploy an AI agent using the latest AI Toolkit and GitHub Copilot features inside VS Code. You'll start with a prototype setup, where Copilot connects to the AI Toolkit through an MCP server to suggest models, open the model playground, instrument code for tracing, enable viewing traces, and finally deploy the agent to Azure. You’ll leave with a functioning, traced, and deployed agent — plus a repeatable local-to-cloud workflow. Join the conversation on the Microsoft Foundry Discord: https://aka.ms/AIDevDays/Discord Sign up for an Azure Free Trial: https://aka.ms/AzureFreeTrialYT Get hands-on with today's MCP lab repo: https://aka.ms/AIDevDays/LabDay1 Catch up on-demand: https://aka.ms/AIDevDays/OnDemand Chapters 00:00 – Welcome & Workshop Overview: AI Toolkit + GitHub Copilot 00:20 – Why Combine AI Toolkit with Copilot? Key Benefits 01:01 – Setting Up Your Environment: Foundry Project & VS Code Extensions 02:32 – Connecting Foundry to VS Code: Default Project Setup 04:41 – GitHub Copilot Agent Mode: Recommended Models & Setup 05:36 – Demo: Getting Model Recommendations with AI Toolkit Tools 09:19 – Testing Models in AI Toolkit Playground: Compare & Choose 12:34 – Building Your Agent: Using Agent Builder in VS Code 14:08 – Creating MCP Servers: Templates & Workflow in AI Toolkit 16:05 – Adding Tools & Improving System Prompts for Your Agent 19:06 – Migrating to Code: Generating Agent Code from Agent Builder 21:19 – Enabling Trac

Published 2025-12-10 · Watch on YouTube ↗
Availability and playback restrictions are controlled by the original channel.