How I’m studying for the GH-600 (GitHub Certified: Agentic AI Developer) — from zero

我是如何从零开始备考 GH-600（GitHub 认证：智能体 AI 开发人员）的

GitHub just launched a new certification — GH-600, “Certified: Agentic AI Developer” — about building and supervising AI agents (like Copilot) inside real projects. I decided to go for it, even though I started with almost no AI background. GitHub 最近推出了一项新认证——GH-600，“认证：智能体 AI 开发人员”（Certified: Agentic AI Developer），内容涵盖如何在实际项目中构建和监管 AI 智能体（如 Copilot）。尽管我几乎没有任何 AI 背景，但我还是决定挑战一下。

The hardest part wasn’t the material. It was knowing what to trust. The official study guide is short, and nearly everything else I found was a site trying to sell me a quiz — with no way to see where their answers came from. So I’m doing it the slow, honest way: reading the official GitHub and Microsoft docs one topic at a time, and learning it by teaching it. 最难的部分并非学习材料本身，而是如何辨别信息的可信度。官方学习指南很简短，而我找到的其他资料几乎都是试图向我兜售模拟题的网站，且无法核实其答案来源。因此，我选择了最稳扎稳打的方式：逐个主题阅读 GitHub 和微软的官方文档，并通过“以教促学”的方式来掌握知识。

As I understand each piece, I write it up as a plain-English lesson — every fact linked back to its official source. Studying and building, in parallel. That turned into a website, and I’ve made it completely free for everyone — no signup, no paywall: agenticlately.com. 每理解一个知识点，我都会将其整理成通俗易懂的教程，并为每个事实标注官方来源链接。我将学习与实践同步进行。最终，这些内容汇集成了一个网站，并且我将其完全免费开放给所有人，无需注册，没有付费墙：agenticlately.com。

It starts from absolute zero, so you can follow even if you’ve never used GitHub or touched AI. I’m being upfront about where it stands: I’ve covered the foundations and the first two areas so far, and I’m adding new lessons as I work through the rest. It’s a journey in progress — if you’re prepping too, follow along and it’ll grow with you. If a lesson doesn’t click, tell me; that’s what makes the next one better. 该教程从零基础开始，即使你从未用过 GitHub 或接触过 AI 也能跟上。我坦诚地说明目前的进度：目前已涵盖基础知识和前两个领域，随着学习的深入，我会持续更新后续课程。这是一个正在进行的过程——如果你也在备考，欢迎加入，它会与你共同成长。如果某个课程让你感到困惑，请告诉我，这能帮助我改进后续的内容。

What the GH-600 actually covers

GH-600 的考试内容概览

For anyone curious, here’s the shape of the exam — six skill areas, with how much each counts (so you know where to spend time): 对于好奇的朋友，以下是考试的结构——包含六个技能领域及其权重（方便你合理分配复习时间）：

#	Domain	Weight
1	Prepare agent architecture and SDLC processes	15–20%
2	Implement tool use and environment interaction	20–25%
3	Manage memory, state, and execution	10–15%
4	Perform evaluation, error analysis, and tuning	15–20%
5	Orchestrate multi-agent coordination	15–20%
6	Implement guardrails and accountability	10–15%

#	领域	权重
1	准备智能体架构和 SDLC 流程	15–20%
2	实现工具使用和环境交互	20–25%
3	管理内存、状态和执行	10–15%
4	执行评估、错误分析和调优	15–20%
5	编排多智能体协作	15–20%
6	实现护栏和问责机制	10–15%

Domain 2 is the heaviest, so I’m front-loading it. Passing score is 700 / 1000, and it’s in beta right now — but the credential is the same as the final version. If you just want the quick overview, I wrote it up here: GH-600 exam guide. 领域 2 的权重最高，所以我优先复习这一部分。及格分数为 700/1000，目前考试处于 Beta 测试阶段，但获得的证书与正式版完全一致。如果你只需要快速概览，我在这里做了总结：GH-600 考试指南。

Good luck if you’re taking it too — I’ll keep posting as the journey continues. 如果你也在备考，祝你好运——我会随着学习进程持续更新。

Independent, unofficial resource. Not affiliated with GitHub or Microsoft. Always confirm details against the official Microsoft Learn study guide. 注：本资源为独立非官方资源，与 GitHub 或微软无关联。请务必以微软官方 Learn 学习指南为准核实细节。