Developing in Agentic AI Systems Study Guide

You are in the right place to prepare for GH-600, Developing in Agentic AI Systems. This guide is the written companion to the practice exams: a complete walk through everything the test covers, gathered in one place.

The exam rewards judgment far more than recall. It hands you a scenario, usually an agent doing real work through GitHub Copilot, and asks which decision settles it. The same instincts win again and again: a control that is enforced beats an instruction the agent is merely asked to follow, the narrowest grant that still meets the requirement beats a broad one, and evidence you can open beats the agent's own report of success. Expect choices such as a line in an instructions file versus a required review or a scoped tools list, or rerunning only the failed jobs versus replaying the whole run.

The guide follows the six official domains, weighted the way the real exam weights them, so tool use and environment interaction gets the most room. Each chapter builds the mental model in plain language first, then separates the look-alike options with comparison tables and decision trees, and most close by reasoning through exam-style scenarios. Start at the top, or pick a domain from the list beside this page.