Certified Tester Testing with Generative AI (CT-GenAI v1.1) — Chapter 2: Prompt Engineering for Effective Software Testing

Prompt engineering is where this syllabus stops describing the technology and starts asking you to use it. The material covers the parts a usable prompt is built from — role, context, the task itself, constraints, and the shape the output should take — and the named techniques: zero-shot, few-shot, chain-of-thought, role prompting, and iterative refinement when the first answer misses. All of it is then pointed at real testing activities: deriving test conditions from a requirement, designing cases from a specification, generating test data, drafting review checklists and summarising defect reports.

This is the chapter that decides most results. It supplies 11 of the 40 questions but 16 of the 46 points, because five of the exam's six K3 questions sit here and a K3 answer is worth two points. Six questions are K2 and the remaining five ask you to apply something: given a test activity and a situation, choose the technique and the prompt structure that fit, or spot why a prompt that was written returns output nobody can use.

Recognition is not enough at this level. Being able to say what few-shot prompting means answers a K2 question; a K3 question hands you a requirement and expects you to know that examples beat explanation when the output format matters, or that a prompt producing fluent but untraceable test conditions needs a constraint on coverage, not a politer request. The only preparation that works is writing prompts against your own backlog and reading what comes back critically.

One number is worth remembering while you practise here: strip Chapter 2 out of a perfect paper and 30 points remain, which is exactly the pass mark. There is no version of this exam that you pass comfortably without this chapter, and the syllabus agrees — 365 of its 815 minutes of required training are spent on it.