GitHub · GH-600
Failure analysis and tuning
GH-600 · Lesson 11 of 16 · 1:29:31
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the GH-600 exam (GitHub Certified: Agentic AI Developer). In this lesson: finding failures in workflow logs, plans, session logs, outputs, and artifacts, classifying root causes as reasoning errors, tool misuse, or context and environment issues, and tuning instructions, workflows, constraints, memory, and tool access, one change at a time. Includes three original practice questions.
Preview the topics
- Contoso’s agent wrote correct code, but its tests never ran, because the workflow it relied on triggers only on pushes to main. How should this failure be classified?
- Contoso’s agent keeps formatting monetary amounts as decimals, although the team stores them as integer cents. What is the best first fix?
- Contoso’s cloud agent cannot download packages from the company’s private registry, and the session logs show a blocked request. What should the team do?