Question 310
A task requests exhaustive detail and a strict fifty-word limit. What should prompt design clarify?
- Add another equally conflicting length instruction.
- Assume the model can fit unlimited detail into any length.
- Repeat both requirements in uppercase.
- Which constraint takes priority and what information must be preserved.
Explanation
Correct Answer
D:Explain which constraint takes priority and how to summarize when both cannot be satisfied, preventing incompatible instructions from remaining unresolved.
Incorrect Answers
A:Additional contradiction makes compliance less determinate.
B:A finite output budget imposes real information tradeoffs.
C:Emphasis does not resolve their practical conflict.
Question 311
A prompt only says “do not be vague.” What addition makes the desired behavior more actionable?
- Require more adjectives.
- Repeat the prohibition many times.
- Remove the task context.
- Define the required evidence and decision criteria.
Explanation
Correct Answer
D:Positive operational guidance defines what success looks like instead of merely naming an undesirable quality.
Incorrect Answers
A:Adjectives do not necessarily supply concrete evidence or criteria.
B:Repetition adds emphasis without defining the missing behavior.
C:Less context makes the desired specificity harder to achieve.
Question 312
A prompt tells Claude “you are the system administrator.” What does that establish?
- Permission to ignore every user constraint.
- Guaranteed correctness of administrative advice.
- Automatic root access to the host.
- A behavioral role, not actual operating-system or service privileges.
Explanation
Correct Answer
D:A role prompt influences framing and behavior; executable privileges still come from the host, credentials, and enforced permission configuration.
Incorrect Answers
A:A role description does not authorize unrelated actions.
B:Persona does not prove factual or operational accuracy.
C:Real privileges depend on the execution environment.
Question 313
A prompt embeds a customer document beside instructions. Why use clear content boundaries or labels?
- To encrypt the document.
- To distinguish material to analyze from directions governing the task.
- To remove the need for tool permissions.
- To guarantee all embedded instructions become trusted.
Explanation
Correct Answer
B:Structure helps interpretation, although delimiters alone are not a complete security boundary against malicious content.
Incorrect Answers
A:Text delimiters are not encryption.
C:Prompt organization does not enforce backend authorization.
D:Document content remains data and does not acquire governing authority.
Question 314
A first model call extracts facts and a second writes a report. What should the handoff contain?
- Only the first model’s confidence adjective.
- The entire transcript with no identified extraction result.
- Invented values for every missing field.
- Validated extracted facts with source references and explicit unknowns.
Explanation
Correct Answer
D:The second stage needs a dependable intermediate contract rather than unsupported conclusions or hidden uncertainty.
Incorrect Answers
A:That omits the facts the report must use.
B:Unstructured handoff makes the relevant intermediate evidence harder to locate.
C:Fabrication contaminates the next stage’s reasoning.
Question 315
A technical explanation is intended for new developers but repeatedly uses unexplained internal acronyms. What prompt adjustment fits?
- Remove every technical detail.
- Specify the audience’s assumed knowledge and require necessary terms to be explained.
- Ask for more acronyms to sound authoritative.
- Change the factual answer to suit the audience.
Explanation
Correct Answer
B:Specify assumed background and require unfamiliar terms to be explained while preserving the underlying facts and technical relationships.
Incorrect Answers
A:Oversimplification can make the explanation incomplete rather than accessible.
C:That increases the barrier the task needs to reduce.
D:Presentation should adapt while accuracy remains intact.
Question 316
A prompt says an account qualifies if it has "identity and address checks," but examples disagree about whether one passed check is sufficient. The business requires both. What clarification is needed?
- State that either check passing is enough.
- Ask for a more confident eligibility label without changing the rule.
- State the AND condition and cover the one-pass/one-fail boundary.
- Remove the distinction between the two checks.
Explanation
Correct Answer
C:Explicitly requiring both checks removes the quantifier ambiguity. A boundary example demonstrates that passing only one does not satisfy the rule.
Incorrect Answers
A:That changes the business rule from conjunction to disjunction and would accept accounts the requirement excludes.
B:Confidence wording cannot resolve a logically ambiguous condition or repair contradictory demonstrations.
D:Collapsing the inputs loses the information required to determine whether both independent prerequisites passed.
Question 317
An extraction prompt includes an example customer named Maria. For records with no customer name, the model repeatedly outputs Maria. What prompt revision addresses this failure?
- Separate example values from current-record evidence.
- Add Maria to every output schema as a default.
- Repeat the example name more often for consistency.
- Remove the current record and provide only the example.
Explanation
Correct Answer
A:The failure is contamination from sample values; the prompt must separate illustrative output from evidence in the live input.
Incorrect Answers
B:A default copied from an example fabricates an identity rather than handling missing evidence.
C:Further repetition can reinforce the incorrect transfer of sample content into unrelated records.
D:Without the actual input, the model cannot perform record-specific extraction at all.
Question 318
A developer updates Claude Code’s local CLAUDE.md but a separate direct API service behaves unchanged. Why?
- The API key permanently stores all local files.
- The API service must explicitly supply its intended instructions in its requests.
- The instruction file changes only network speed.
- Every API model ignores system instructions.
Explanation
Correct Answer
B:A direct API service receives the inputs its code sends; it does not automatically load another product's local project instruction files.
Incorrect Answers
A:Credentials do not automatically synchronize repository instructions.
C:Its purpose is behavioral context, not transport performance.
D:The issue is missing transmission, not universal instruction rejection.
Question 319
A prompt fails only on a known abbreviation. What is a useful first revision strategy?
- Rewrite every instruction and change models simultaneously.
- Remove all successful examples.
- Rewrite the entire instruction set and all examples before isolating the abbreviation.
- Clarify the abbreviation and run regression checks.
Explanation
Correct Answer
D:A focused modification makes its effects easier to assess while checking for unintended changes elsewhere. Add the abbreviation's intended meaning and evaluate the affected cases plus existing regressions before considering broader prompt changes.
Incorrect Answers
A:That makes the cause of improvement or regression harder to identify.
B:Unnecessary deletion can damage behavior unrelated to the observed issue.
C:Broad simultaneous changes make it harder to determine whether the targeted ambiguity was resolved without causing unrelated regressions.
Question 320
An answer has perfect headings but omits a required comparison. What should evaluation and prompt refinement prioritize?
- Mark success because every heading exists.
- Apply a stricter heading-style rule before checking missing information.
- The missing substantive requirement, not formatting alone.
- Score only the sections with content and exclude omissions from the average.
Explanation
Correct Answer
C:The acceptance contract includes substantive comparison content, so cosmetic compliance must not substitute for checking the required decision dimensions.
Incorrect Answers
A:Structural appearance is only part of the acceptance contract.
B:Additional style enforcement does not address the demonstrated omission. The required comparison needs a substantive acceptance check and prompt correction.
D:Excluding a required absent comparison hides a task failure and inflates the score. Evaluation must account for required content that was not produced.