Module 1: Explore Microsoft Copilot Studio and agent harnesses
Microsoft Copilot Studio supports agents and workflows with different runtime architectures. This module explains how harness choice affects orchestration, available capabilities, authoring, publishing, and billing. You compare the GitHub Copilot, standard, and Copilot Chat harnesses and select a harness for a business scenario.
Module 2: Create instruction-driven agents in Microsoft Copilot Studio
Agents powered by the GitHub Copilot harness use natural-language instructions as the primary way to define their behavior. This module explains how to define an agent's purpose, scope, behavior, boundaries, and grounding, and then refine the agent by testing it in Preview.
Module 3: Ground and extend agents with knowledge, tools, and skills
Agents powered by the GitHub Copilot harness combine models, knowledge, tools, skills, and connected agents to complete tasks by using approved information and current data. This module explains how to configure core behavior and these building blocks, validate their use in Preview, and refine the descriptions that help the agent choose what to use.
Module 4: Automate agent tasks with workflows in Microsoft Copilot Studio
Workflows give agents a deterministic, auditable path for actions that must produce the same result every time. This module explains when to use a workflow instead of agent reasoning, how to design a clear input and output contract, how to build and publish a standalone workflow, and how to add it to an agent with end-user confirmation and complete test coverage.
Module 5: Evaluate, publish, and manage agents in Microsoft Copilot Studio
Moving an agent from the Preview tab into production requires repeatable quality checks, controlled publication, post-publish validation, and ongoing operational oversight. This module explains how to create and run evaluations, validate connected-agent delegation, interpret quality and runtime signals, publish an agent to Teams and Microsoft Copilot, and identify when governance controls, application lifecycle management practices, or administrator support are required.