3.1 Structured Prompting & Six Professional Scrum Master Prompts

3.1 Structured Prompting and Six Professional Scrum Master Prompts

Core principle: Effective prompting is deliberate communication with a probabilistic tool. Supply enough trusted context to do a bounded task, request a useful form of output, and inspect the result. No prompt structure makes the model authoritative or transfers a Scrum accountability.

What the Official Preparation Page Establishes

Effective AI Prompting is one of four published PSM-AI Essentials question categories. Scrum.org's preparation page currently curates three resources for it: 6 Professional AI Prompts for Scrum Masters, an article about context engineering, and a rapid-prototyping example. It does not designate CRAFTSS, APE, CO-STAR, RACE, or another lettered formula as the official exam framework. Those structures may be useful elsewhere, but candidates should not turn one into an invented Scrum.org rule.

A robust prompt can be built without an acronym:

A role instruction such as “offer the perspective of an experienced facilitator” can focus vocabulary and questions. It does not give the model real credentials, privileged knowledge, or reliable judgment. Source material and evaluation carry more evidentiary weight than a persona.

Six Curated Scrum Master Uses

Scrum.org's six-prompt article demonstrates prompting for:

1. Sprint Goal creation — synthesize a coherent objective from Product Goal context and the work being considered; the Scrum Team still defines the Sprint Goal collaboratively.
2. Product Goal creation — draft a measurable, understandable long-term objective for Product Owner and stakeholder inspection; the Product Owner remains accountable for developing and explicitly communicating it.
3. Removing an impediment — generate causes, stakeholder questions, and small experiments instead of pretending the first proposed fix is proven.
4. Definition of Done discussion — examine process, technical, delivery, industry, organizational, and non-functional expectations, then add AI-impact expectations where relevant. Developers remain accountable for quality and adherence.
5. DORA improvement ideas — use verified delivery measures as signals, look for system constraints, and avoid ranking individuals or treating a metric as a target.
6. Conflict resolution — prepare neutral questions and alternative interpretations while preserving direct human conversation, consent, and psychological safety.

The published prompts are examples of professional use, not new Scrum events or mandatory templates. FOCUS and SMART appear as useful structuring devices in examples; they do not supersede the Scrum Guide.

Worked Example: Preparing a Conflict Conversation

Suppose a Scrum Master hears two Developers describe the same design disagreement in incompatible terms. The goal is not to ask a model to decide who is right. A bounded prompt could say:

TASK
Using the sanitized observations below, draft neutral questions that could help the
two people surface assumptions and interests in a direct conversation.

CONTEXT
- One person prefers a small change that preserves the current service.
- One person prefers replacement because of recurring support costs.
- Both agree on the Sprint Goal; cost evidence is incomplete.

CONSTRAINTS
- Do not infer motives, emotion, seniority, or blame.
- Do not recommend a winner.
- Label statements that need evidence.
- Do not include names or private transcript excerpts.

OUTPUT
Return: (1) six neutral questions, (2) evidence the team may need, and
(3) two small experiments. Stop there for human review.

The Scrum Master checks every question for loaded language and decides whether AI use is appropriate to disclose. The people then have the conversation themselves. If the prompt contains invented motives, the Scrum Master changes the context and constraints or discards the output.

Inspect More Than Style

A polished answer is not necessarily a good answer. Evaluate whether it addressed the requested task, stayed within supplied facts, exposed important unknowns, avoided unsafe data, respected Scrum accountabilities, and can be tested or corroborated. For a Sprint Goal draft, the proof is not that the sentence sounds inspiring; it is whether the Scrum Team can use it as a coherent objective for the Sprint. For an impediment idea, the proof is an observed result from a safe experiment. This is prompt engineering used inside empiricism: transparency about inputs, inspection of evidence, and adaptation based on what actually happens.