After this day you will be able to run AI agents in your development process and trace what they do at any time. You will separate planning from execution, connect tools and MCP servers with clear permissions, keep state and context stable across long tasks and find the cause when a run fails. For several agents you will define isolation, handoffs and recovery. The course is aimed at experienced developers and platform and DevOps engineers who use coding agents with GitHub as the control plane and are preparing for the GH-600 exam.
What you’ll take away.
- Integrate agents into the development lifecycle so that planning, approval and execution stay separate
- Select tools and MCP servers for agents, set permissions and maintain allow lists
- Manage agent memory and state so that long-running tasks do not drift
- Trace failures to their cause using logs, plans and artefacts, and tune agent behaviour
- Set autonomy levels by risk and add human approvals where they actually reduce risk
Your training in detail
- Who is it for?
- Experienced developers, platform and DevOps engineers and architects who run and supervise coding agents such as GitHub Copilot in production team workflows.
- Prerequisites
- You know the software development lifecycle, GitHub workflows, branch protection and code reviews, and you have already worked with GitHub Copilot or other coding agents. Experience with custom instructions or MCP servers helps.
- Format & planning
- Standard length: 1 training day. A different length can be arranged for companies.
Agenda: from understanding to application
Our independently developed agenda proposal. Focus areas, lab environment and scope are agreed before booking.
Agent architecture in the SDLC: tasks, inputs and outputs, success criteria and anti-patterns
Tools and MCP: the GitHub remote MCP server, registries, allow lists and permissions
Execution in the repository and CI: scope, branches, pull requests, retries and rollbacks
Memory and state: short- and long-term memory, durable artefacts, stale context
Evaluation and tuning: success signals, scan results and root-cause analysis
Multi-agent orchestration: isolation, conflicts, handoffs and recovery
Guardrails: autonomy levels, least privilege and approvals for irreversible changes
In a practice repository you have an agent write a structured plan to a file first and only open a branch and pull request after your approval. You connect an MCP server through an allow list, trigger a failing run and classify the cause from the log and artefacts: a reasoning error, a misused tool or missing context.
Exams, licences & credentials
The course prepares you for the GH-600 exam, which leads to the GitHub Certified: Agentic AI Developer certification. The exam and its fee are included only if the offer explicitly confirms this. The certificate is issued by GitHub through Microsoft Learn. This is not an official GitHub or Microsoft course unless the offer states otherwise.
Vendor references & review date
Sources checked on 2026-11-10. These sources describe the technology or external exam requirements. They do not imply a partnership with NextNowa.
Your learning content
These objectives describe the training topic. The specific syllabus, dates and trainer are agreed before booking; the written offer is binding.
Frequently asked questions
Which certification does GH-600 lead to?
According to Microsoft Learn, passing GH-600 earns the GitHub Certified: Agentic AI Developer certification. The exam lasts 120 minutes and is currently offered in English.
How does GH-600 differ from GH-300?
GH-300 covers GitHub Copilot as a coding assistant. GH-600 goes further and focuses on operating, supervising and safeguarding autonomous agents and multi-agent workflows in the development lifecycle.
Is one day enough to prepare?
The day brings the six exam domains together and practises the key workflows. The exam profile assumes hands-on experience with coding agents, which you should build up in your own repositories as well.
Does GH-600 Developing in Agentic AI Systems: governing AI agents in the development lifecycle include a certificate?
The course prepares you for the GH-600 exam, which leads to the GitHub Certified: Agentic AI Developer certification. The exam and its fee are included only if the offer explicitly confirms this. The certificate is issued by GitHub through Microsoft Learn. This is not an official GitHub or Microsoft course unless the offer states otherwise.