WGU D685 Practical Applications of Prompt
Engineering OA2 Practice Exam ACTUAL
EXAM 2026/2027 | WGU D685 Prompt
Engineering | Verified Q&A | Pass
Guaranteed - A+ Graded
Section 1: Foundational Prompt Engineering Techniques (15 Questions)
Q1: A student submits the following zero-shot prompt to a language model: "Write a business email."
The model generates a generic, unfocused message. Which adjustment most improves this prompt
using foundational prompt engineering principles?
A. Add more adjectives like "professional" and "concise"
B. Specify role, audience, purpose, tone, and key points to cover [CORRECT]
C. Increase the temperature setting to 0.9
D. Submit the same prompt three times and average the outputs
Correct Answer: B
Rationale: The original prompt fails on multiple WGU D685 rubric criteria: no role definition (who is
writing), no audience specification (who receives it), no purpose (why), no tone (how), and no content
requirements (what). Option B addresses all five elements of effective prompt structure. Option A adds
vague descriptors without structural clarity. Option C increases randomness, worsening consistency.
Option D is computationally wasteful and doesn't fix the underlying prompt deficiency.
Q2: A data analyst needs to classify customer support tickets into categories: Billing, Technical, Account,
or General. They provide this prompt with three examples of each category, then ask the model to
classify a new ticket. This technique is best described as:
A. Zero-shot prompting
B. Few-shot prompting [CORRECT]
C. Chain-of-thought prompting
D. Fine-tuning
,Correct Answer: B
Rationale: Few-shot prompting provides exemplars (input-output pairs) within the context window to
guide the model's pattern recognition without parameter updates. The analyst included three examples
per category—classic few-shot structure per WGU D685 Module 2. Zero-shot (A) provides no examples.
Chain-of-thought (C) requires reasoning steps in exemplars. Fine-tuning (D) involves gradient descent
training, not in-context learning.
Q3: A developer is designing a prompt for a medical coding assistant. The system instruction states: "You
are a certified medical coder with 10 years of experience. Use ICD-10-CM guidelines. When uncertain,
ask clarifying questions rather than guessing." This design primarily leverages which prompt engineering
component?
A. Temperature manipulation
B. Role-based system prompting with explicit behavioral constraints [CORRECT]
C. Top-p nucleus sampling
D. Token length optimization
Correct Answer: B
Rationale: The prompt establishes a specific role (certified medical coder), domain expertise (10 years),
knowledge constraints (ICD-10-CM), and behavioral guardrails (clarify vs. guess)—exemplifying role-
based system prompting with constraint application. Temperature (A) and top-p (C) are inference
parameters, not prompt content. Token optimization (D) is unrelated to this structure.
Q4: A prompt engineer wants a model to solve a complex multi-step math problem by showing its
reasoning before giving the final answer. Which technique should they use?
A. Zero-shot prompting with temperature 0
B. Chain-of-thought prompting with explicit reasoning request [CORRECT]
C. Few-shot prompting with only final answers in examples
D. Tree-of-thoughts with parallel reasoning paths
Correct Answer: B
Rationale: Chain-of-thought (CoT) prompting explicitly requests intermediate reasoning steps (e.g.,
"Let's think step by step" or "Show your work"), improving accuracy on complex reasoning tasks by 40-
100% per WGU D685 research citations. Zero-shot (A) without CoT trigger underperforms. Few-shot
with only answers (C) misses the reasoning demonstration. Tree-of-thoughts (D) is over-engineered for a
single linear math problem.
, Q5: [SELECT-ALL-THAT-APPLY] Which elements should be included in a well-structured system prompt
for a healthcare documentation assistant? (Select all that apply.)
A. Role definition and domain expertise [CORRECT]
B. Output format specifications (e.g., SOAP note structure) [CORRECT]
C. Prohibited content and safety guardrails [CORRECT]
D. The user's specific medical history for personalization
Correct Answer: A, B, C
Rationale: Per WGU D685 Module 1, effective system prompts require: (A) role definition to activate
relevant domain knowledge, (B) output format specifications to ensure structured, parseable responses,
and (C) safety guardrails to prevent harmful outputs. Option D is incorrect because system prompts
should contain universal instructions applicable to all users; personal medical history belongs in user
prompts to maintain privacy and context separation.
Q6: A model generates inconsistent outputs for the same prompt across multiple runs. Which inference
parameter adjustment would most improve output determinism?
A. Increase temperature to 1.0
B. Set temperature to 0.0 and use fixed seed if available [CORRECT]
C. Increase top-p to 0.95
D. Decrease max_tokens to 50
Correct Answer: B
Rationale: Temperature controls randomness in token sampling: 0.0 = deterministic (greedy decoding),
higher values = more random. For consistent outputs, temperature 0.0 is essential per WGU D685
inference parameter guidelines. Increasing temperature (A) or top-p (C) increases variability.
Max_tokens (D) truncates output length without affecting consistency.
Q7: A prompt engineer is designing a tree-of-thoughts approach for strategic business planning. The
model generates multiple reasoning paths, evaluates each against criteria, then selects the best. What is
the primary advantage of this technique over standard chain-of-thought?
A. Reduced token consumption
B. Exploration of multiple hypotheses with self-evaluation and backtracking [CORRECT]
C. Faster inference speed
D. Elimination of hallucinations
Engineering OA2 Practice Exam ACTUAL
EXAM 2026/2027 | WGU D685 Prompt
Engineering | Verified Q&A | Pass
Guaranteed - A+ Graded
Section 1: Foundational Prompt Engineering Techniques (15 Questions)
Q1: A student submits the following zero-shot prompt to a language model: "Write a business email."
The model generates a generic, unfocused message. Which adjustment most improves this prompt
using foundational prompt engineering principles?
A. Add more adjectives like "professional" and "concise"
B. Specify role, audience, purpose, tone, and key points to cover [CORRECT]
C. Increase the temperature setting to 0.9
D. Submit the same prompt three times and average the outputs
Correct Answer: B
Rationale: The original prompt fails on multiple WGU D685 rubric criteria: no role definition (who is
writing), no audience specification (who receives it), no purpose (why), no tone (how), and no content
requirements (what). Option B addresses all five elements of effective prompt structure. Option A adds
vague descriptors without structural clarity. Option C increases randomness, worsening consistency.
Option D is computationally wasteful and doesn't fix the underlying prompt deficiency.
Q2: A data analyst needs to classify customer support tickets into categories: Billing, Technical, Account,
or General. They provide this prompt with three examples of each category, then ask the model to
classify a new ticket. This technique is best described as:
A. Zero-shot prompting
B. Few-shot prompting [CORRECT]
C. Chain-of-thought prompting
D. Fine-tuning
,Correct Answer: B
Rationale: Few-shot prompting provides exemplars (input-output pairs) within the context window to
guide the model's pattern recognition without parameter updates. The analyst included three examples
per category—classic few-shot structure per WGU D685 Module 2. Zero-shot (A) provides no examples.
Chain-of-thought (C) requires reasoning steps in exemplars. Fine-tuning (D) involves gradient descent
training, not in-context learning.
Q3: A developer is designing a prompt for a medical coding assistant. The system instruction states: "You
are a certified medical coder with 10 years of experience. Use ICD-10-CM guidelines. When uncertain,
ask clarifying questions rather than guessing." This design primarily leverages which prompt engineering
component?
A. Temperature manipulation
B. Role-based system prompting with explicit behavioral constraints [CORRECT]
C. Top-p nucleus sampling
D. Token length optimization
Correct Answer: B
Rationale: The prompt establishes a specific role (certified medical coder), domain expertise (10 years),
knowledge constraints (ICD-10-CM), and behavioral guardrails (clarify vs. guess)—exemplifying role-
based system prompting with constraint application. Temperature (A) and top-p (C) are inference
parameters, not prompt content. Token optimization (D) is unrelated to this structure.
Q4: A prompt engineer wants a model to solve a complex multi-step math problem by showing its
reasoning before giving the final answer. Which technique should they use?
A. Zero-shot prompting with temperature 0
B. Chain-of-thought prompting with explicit reasoning request [CORRECT]
C. Few-shot prompting with only final answers in examples
D. Tree-of-thoughts with parallel reasoning paths
Correct Answer: B
Rationale: Chain-of-thought (CoT) prompting explicitly requests intermediate reasoning steps (e.g.,
"Let's think step by step" or "Show your work"), improving accuracy on complex reasoning tasks by 40-
100% per WGU D685 research citations. Zero-shot (A) without CoT trigger underperforms. Few-shot
with only answers (C) misses the reasoning demonstration. Tree-of-thoughts (D) is over-engineered for a
single linear math problem.
, Q5: [SELECT-ALL-THAT-APPLY] Which elements should be included in a well-structured system prompt
for a healthcare documentation assistant? (Select all that apply.)
A. Role definition and domain expertise [CORRECT]
B. Output format specifications (e.g., SOAP note structure) [CORRECT]
C. Prohibited content and safety guardrails [CORRECT]
D. The user's specific medical history for personalization
Correct Answer: A, B, C
Rationale: Per WGU D685 Module 1, effective system prompts require: (A) role definition to activate
relevant domain knowledge, (B) output format specifications to ensure structured, parseable responses,
and (C) safety guardrails to prevent harmful outputs. Option D is incorrect because system prompts
should contain universal instructions applicable to all users; personal medical history belongs in user
prompts to maintain privacy and context separation.
Q6: A model generates inconsistent outputs for the same prompt across multiple runs. Which inference
parameter adjustment would most improve output determinism?
A. Increase temperature to 1.0
B. Set temperature to 0.0 and use fixed seed if available [CORRECT]
C. Increase top-p to 0.95
D. Decrease max_tokens to 50
Correct Answer: B
Rationale: Temperature controls randomness in token sampling: 0.0 = deterministic (greedy decoding),
higher values = more random. For consistent outputs, temperature 0.0 is essential per WGU D685
inference parameter guidelines. Increasing temperature (A) or top-p (C) increases variability.
Max_tokens (D) truncates output length without affecting consistency.
Q7: A prompt engineer is designing a tree-of-thoughts approach for strategic business planning. The
model generates multiple reasoning paths, evaluates each against criteria, then selects the best. What is
the primary advantage of this technique over standard chain-of-thought?
A. Reduced token consumption
B. Exploration of multiple hypotheses with self-evaluation and backtracking [CORRECT]
C. Faster inference speed
D. Elimination of hallucinations
