GH-600T00: Developing in Agentic AI Systems Lab

This lab includes MS Learn content which is fully integrated in the account portal and lab environment.

This lab teaches developers how to build and coordinate agentic AI systems on GitHub, moving from single agent automation to a multi-agent workflow that plans, designs, codes, and validates a small application. Learners work entirely inside GitHub Codespaces with Visual Studio Code, the GitHub agentic workflows CLI extension (gh aw), GitHub Copilot Chat, and GitHub Copilot CLI, authoring agent and workflow definitions as markdown files under .github/ and compiling them into hardened GitHub Actions workflows. Module 1 covers tooling, Model Context Protocol configuration, and agent execution environments, including repository secrets, Actions permissions, and safe-outputs; Module 2 covers multi-agent orchestration with a prebuilt Orchestrator, Planner, Designer, and Coder team. Security and verification run through both modules: fine-grained personal access tokens with least-privilege scope, secrets added only through the repository secrets interface, workflows that are read-only by default, and every agent-authored change delivered as a pull request or documented handoff for human review. Both modules are exercise driven, so learners finish with a working automated workflow, a runnable dashboard, and the documentation trail that proves each agent did its part.

Labs are available in English.