GitHub Certified: Agentic AI Developer credential artwork

GitHub Certified: Agentic AI Developer (GH-600) Courses

Validates knowledge of running AI coding agents in GitHub workflows, including tool use, state, evaluation, orchestration and guardrails.

GH-600 tests how you run AI coding agents inside GitHub workflows, so a good course covers agent design, tools and MCP servers, and guardrails. We compared the official courses on the domains they cover, depth and price.

Take all three free modules in order, starting with Foundations of Agentic AI in GitHub. Skip Tooling, MCP, and Agent Execution Environments only if you already configure MCP servers, and build real agent workflows alongside them.

CourseBest forFormatPrice
Foundations of Agentic AI in GitHubOfficial starting pointInteractiveFree
Designing Agent Architecture and SDLC IntegrationAgent design and SDLCInteractiveFree
Tooling, MCP, and Agent Execution EnvironmentsMCP and agent toolingInteractiveFree
Official starting point

Foundations of Agentic AI in GitHub

The official introduction to the plan, act and evaluate agent lifecycle and to GitHub's role in assigning, reviewing and merging agent work. The GH-600 study guide recommends it as a starting point.

Great fit if you want the official starting point

Poor fit if you need the later domains; continue with the next two modules

Agent design and SDLC

Designing Agent Architecture and SDLC Integration

The official module on designing agent architecture and fitting agents into the software development lifecycle. It covers the exam's first domain, which carries 15 to 20 percent of the exam.

Great fit if you are working on the first domain

Poor fit if you want hands-on labs; build agent workflows yourself alongside it

MCP and agent tooling

Tooling, MCP, and Agent Execution Environments

The official module on tools, MCP servers and the environments agents run in. It covers the exam's largest domain, so it deserves the most study time of the three.

Great fit if you are new to MCP servers

Poor fit if you already configure MCP servers and agent setup steps; go to the study guide's linked docs

How to choose a GitHub Certified: Agentic AI Developer course

  • Start from the study guide. The official GH-600 study guide recommends these modules and links documentation for each of the six domains. Use it to decide what to read after the modules.
  • Expect gaps beyond three modules. The modules cover the agent lifecycle, architecture and tooling. The exam also tests memory and state, evaluating agent output, coordinating several agents and guardrails, so fill those from the study guide's linked docs.
  • Give MCP and tooling the most time. Tooling, MCP, and Agent Execution Environments covers the largest domain, which carries 20 to 25 percent of the exam.
  • Build before you test. You will read real agent configuration and logs. Build small agent workflows with Copilot agents, MCP servers and GitHub Actions so the exam material is familiar.

Recommended GitHub Certified: Agentic AI Developer course combinations

  • Lowest cost: The three Microsoft Learn modules with the free GH-600 Study Guide and Mock Exam from jtur671, an open repository.
  • Case-study practice: The three modules with agenticlately's GH-600 Practice Pack ($10), which has free sample questions.
  • Largest question bank: The three modules with Jimmy Hung's GH-600: Developing in Agentic AI Systems - Practice Tests ($35).
  • Already run agents: Skip to the free mock exam to find weak domains, then read only the modules that cover them.

Compare all GitHub Certified: Agentic AI Developer practice tests →

GitHub Certified: Agentic AI Developer course FAQ

What is the best GH-600 course?

The three official Microsoft Learn modules: they are free, and the GH-600 study guide recommends them. Take Foundations of Agentic AI in GitHub first.

Can I pass GH-600 with free resources?

Yes. The modules and the study guide's linked docs are free, and the community mock exam adds practice. Build agent workflows yourself as well.

How long does GH-600 preparation take?

Plan for about 40 hours in total. The modules are only the start, and hands-on work with agents takes most of the time.

Do I need agent experience before GH-600?

No prerequisite is required, but GitHub expects hands-on experience with Copilot agents, MCP servers and custom instructions. Build that alongside the modules.