Microsoft launches agentic AI certification

Microsoft and GitHub have opened a new beta certification for developers working with AI agents inside software delivery pipelines, adding a credential focused on operating, supervising and governing agentic systems rather than just building prompts. The new badge, listed as “GitHub Certified: Agentic AI Developer (beta),” is tied to Exam GH-600 and is hosted across Microsoft Learn and GitHub Learn. Microsoft disclosed the beta in a May 13 post on its Tech Community site, and the certification pages were live as of May 18. The companies describe the role as working with AI agents inside production-grade software development lifecycle workflows, with GitHub serving as the system of record and control plane. ### When did Microsoft and GitHub actually launch this certification? Microsoft published the announcement on May 13, 2026, not May 18, in a Tech Community post introducing “GitHub Certified: Agentic AI Developer” and inviting developers to take the beta exam. The post said the credential is role-based and centered on how developers and teams “operate, supervise, and integrate AI agents” across the software development lifecycle. (techcommunity.microsoft.com) The certification page on Microsoft Learn now lists the exam as in beta and says the test is provided by Microsoft while the exam and certification are maintained by GitHub. GitHub Learn also lists the exam as beta and available for scheduling through Pearson VUE testing centers or online. ### What skills does GH-600 test? Microsoft Learn says the certification is aimed at people with expertise in “operating, integrating, supervising, and governing AI agents” in production SDLC workflows. (techcommunity.microsoft.com) The listed responsibilities include operating agent workflows, supervising autonomous behavior with GitHub controls, evaluating outputs using scans and artifacts, configuring custom agents and coordinating multi-agent execution safely. (learn.microsoft.com) GitHub Learn’s exam outline adds the operational detail that many early social posts summarized only loosely. The outline includes configuring planning separately from execution, requiring human approval before action, setting degrees of autonomy and guardrails, producing inspectable artifacts, handling environment-specific constraints, and managing memory, state and long-running execution. It also lists identifying failures through logs, plans, traces, outputs and workflow artifacts, classifying root causes, and configuring observability for multi-agent behavior. (learn.microsoft.com) ### Does the curriculum really cover reliability, observability and debugging? GitHub Learn’s published outline supports that description. The outline explicitly references safe execution paths, robust error handling, logs, traces, workflow artifacts, evaluation signals, root-cause classification and observability for multi-agent behavior. Those are the closest official matches to the reliability, observability, debugging and trace-capture themes circulating in posts about the launch. (learn.github.com) Microsoft’s training materials use similar language. A Microsoft Learn path called “Developing in Agentic AI Systems Part 1 of 2” says candidates learn to define execution boundaries, improve reliability and control, and configure tools, permissions and MCP servers. Separate Microsoft documentation on AI observability and tracing says developers can monitor AI systems with traces, logs, evaluation metrics and model outputs, and use tracing to capture latency, exceptions, prompt content, tool usage and retries for debugging. (learn.github.com) ### Who is Microsoft targeting with this exam? GitHub Learn names architects, platform engineers, DevOps engineers, application developers, product managers and security engineers as the audience profile. Microsoft’s announcement similarly says the certification could fit software developers, platform engineers, DevOps engineers, security engineers and technical product managers working with AI-assisted or agent-driven development workflows. (learn.microsoft.com) The Microsoft Learn certification page classifies the level as intermediate and lists roles including AI engineer, app maker, data engineer, developer, DevOps engineer and solution architect. It also says candidates should already have experience with GitHub workflows, code quality, security and review practices, as well as coding agents such as GitHub Copilot, MCP servers and agent customization. (learn.github.com) ### What is still missing? GitHub Learn and Microsoft Learn do not present the beta in exactly the same way. GitHub Learn lists the exam duration as 100 minutes and certificate validity as 24 months, while Microsoft Learn says candidates will have 120 minutes to complete the assessment. Microsoft Learn also says beta candidates will not receive results immediately and that scores will be released about eight weeks after the beta period concludes. (learn.microsoft.com) Microsoft’s certification page says “No training available for this exam,” but the study guide and linked learning paths are already published. That suggests the companies have released preparation materials and the beta exam before publishing a fuller final package for general availability. That is an inference based on the current state of the official pages. (learn.github.com) ### Where can developers track the next step? Microsoft Learn hosts the certification page and the GH-600 study guide, while GitHub Learn hosts the exam details and outline. As of May 18, both pages were live, beta scheduling was available, and Microsoft said score reports would come roughly eight weeks after the beta period ends. (learn.microsoft.com)