Create agents with the GitHub Copilot harness in Microsoft Copilot Studio

  • Learning Path
  • 5 Modules

At a glance

Build agents powered by the GitHub Copilot harness in Microsoft Copilot Studio. Explore agent harnesses, define agent behavior with instructions, ground agents with knowledge, extend them with tools and skills, automate tasks with workflows, and prepare agents for production through evaluation, publishing, monitoring, and governance.

Prerequisites

  • Basic understanding of artificial intelligence and agents
  • Access to a Microsoft Copilot Studio environment where you can create, evaluate, and publish agents and workflows
  • Access to Copilot Credits in the Power Platform environment
  • A Microsoft 365 account with an Outlook mailbox and access to Microsoft Teams
  • Permission to install an agent in Microsoft Teams or Microsoft 365 Copilot, or access to an administrator who can approve the agent

Modules in this learning path

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.

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.

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.

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.

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.

Start