Question 310
A GitHub Copilot task needs TWO complementary actions: the immediate fix and a preventive practice. Context: An engineer wants Copilot to learn an organization's uncommon error-reporting style from two known examples. Observed failure: The model uses generic error messages that omit an internal trace ID. Select exactly TWO decisions that together address these needs.
- Include the exact compiler diagnostic, affected file, minimal relevant context and expected behavior.
- Provide two short representative input-output pairs followed by the new task as few-shot context.
- Use an explicit zero-shot description with parameters, expected return type and error behavior.
- Maintain contract examples for every nontrivial data-format request.
- Maintain reviewed few-shot examples and update them with the contract version.
Explanation
Correct Answers
B:The organization-specific target pattern was not demonstrated in the prompt. The needed intervention is provide two short representative input-output pairs followed by the new task as few-shot context.
E:The organization-specific target pattern was not demonstrated in the prompt. Maintain reviewed few-shot examples and update them with the contract version.
Incorrect Answers
A:The necessary failure evidence was missing from the prompt. The organization-specific target pattern was not demonstrated in the prompt.
C:The task contract was too vague even for a zero-shot request. The organization-specific target pattern was not demonstrated in the prompt.
D:If you maintain contract examples for every nontrivial data-format request, you still need to provide two short representative input-output pairs followed by the new task as few-shot context.
Question 311
The current design is: Provide two short representative input-output pairs followed by the new task as few-shot context. The prior failure was: The model uses generic error messages that omit an internal trace ID. Which ONE ongoing control best protects against recurrence of that specific failure?
- Maintain contract examples for every nontrivial data-format request.
- Include error text and reproducible commands when requesting repairs.
- Maintain reviewed few-shot examples and update them with the contract version.
- Add parameter and return examples when a concise zero-shot prompt fails.
Explanation
Correct Answer
C:The organization-specific target pattern was not demonstrated in the prompt. Maintain reviewed few-shot examples and update them with the contract version. Preventing this failure involves maintaining reviewed few-shot examples and update them with the contract version.
Incorrect Answers
A:Maintain contract examples for every nontrivial data-format request. It would leave tested AI-assisted changes that preserve stated requirements without the specified safeguard.
B:Shell commands from AI tools must be inspected for resolved paths and side effects before destructive execution. Preventing this failure involves maintaining reviewed few-shot examples and update them with the contract version.
D:Add parameter and return examples when a concise zero-shot prompt fails. It would leave tested AI-assisted changes that preserve stated requirements without the specified safeguard.
Question 312
A team reports: A junior developer has no sample input or output and asks Copilot for a high-level function stub. They observe: A generated stub returns an untyped dictionary where an integer is required. Which ONE technical decision most directly corrects the cause?
- Use an explicit zero-shot description with parameters, expected return type and error behavior.
- Include the required CSV dialect, escaping rules and example input-output records in the prompt.
- Include the exact compiler diagnostic, affected file, minimal relevant context and expected behavior.
- Provide two short representative input-output pairs followed by the new task as few-shot context.
Explanation
Correct Answer
A:The task contract was too vague even for a zero-shot request.
Incorrect Answers
B:Including the required CSV dialect, escaping rules and example input-output records in the prompt would not meet the need for an explicit zero-shot description with parameters, expected return type and error behavior.
C:Including the exact compiler diagnostic, affected file, minimal relevant context and expected behavior would not meet the need for an explicit zero-shot description with parameters, expected return type and error behavior.
D:Providing two short representative input-output pairs followed by the new task as few-shot context would not meet the need for an explicit zero-shot description with parameters, expected return type and error behavior.
Question 313
A team has implemented the following change: Use an explicit zero-shot description with parameters, expected return type and error behavior. The original requirement was: A junior developer has no sample input or output and asks Copilot for a high-level function stub. Which ONE validation test best demonstrates the intended behavior?
- Check that the generated signature and error branch match the verbal specification.
- Ask for a third error example and compare the field names, tone and structure to the references.
- Test a sample containing embedded commas, doubled quotes and line breaks.
- Rerun the compiler and confirm the specific diagnostic is gone without new warnings.
Explanation
Correct Answer
A:Check that the generated signature and error branch match the verbal specification.
Incorrect Answers
B:Ask for a third error example and compare the field names, tone and structure to the references. That check cannot demonstrate compatibility with the pinned dependency version.
C:Generated code requires independently defined acceptance tests and review against the real program contract. Verification of this result involves checking that the generated signature and error branch match the verbal specification.
D:Rerun the compiler and confirm the specific diagnostic is gone without new warnings. Verification of this result involves checking that the generated signature and error branch match the verbal specification.
Question 314
During troubleshooting, investigators observe: A generated stub returns an untyped dictionary where an integer is required. The system context is: A junior developer has no sample input or output and asks Copilot for a high-level function stub. Which ONE cause most directly explains the observed failure?
- The necessary failure evidence was missing from the prompt.
- The task contract was too vague even for a zero-shot request.
- Ambiguous format requirements left the model to invent unsupported assumptions.
- The organization-specific target pattern was not demonstrated in the prompt.
Explanation
Correct Answer
B:The remedy should therefore be: Use an explicit zero-shot description with parameters, expected return type and error behavior.
Incorrect Answers
A:The necessary failure evidence was missing from the prompt. The task contract was too vague even for a zero-shot request.
C:Ambiguous format requirements left the model to invent unsupported assumptions. The task contract was too vague even for a zero-shot request.
D:The organization-specific target pattern was not demonstrated in the prompt. The missing change involves the task contract was too vague even for a zero-shot request.
Question 315
The current design is: Use an explicit zero-shot description with parameters, expected return type and error behavior. The prior failure was: A generated stub returns an untyped dictionary where an integer is required. Which ONE ongoing control best protects against recurrence of that specific failure?
- Maintain reviewed few-shot examples and update them with the contract version.
- Include error text and reproducible commands when requesting repairs.
- Maintain contract examples for every nontrivial data-format request.
- Add parameter and return examples when a concise zero-shot prompt fails.
Explanation
Correct Answer
D:The task contract was too vague even for a zero-shot request. Add parameter and return examples when a concise zero-shot prompt fails.
Incorrect Answers
A:Generated code requires independently defined acceptance tests and review against the real program contract. It would leave add parameter and return examples when a concise zero-shot prompt fails without the specified safeguard.
B:Shell commands from AI tools must be inspected for resolved paths and side effects before destructive execution. It would leave add parameter and return examples when a concise zero-shot prompt fails without the specified safeguard.
C:Maintain contract examples for every nontrivial data-format request. It would leave add parameter and return examples when a concise zero-shot prompt fails without the specified safeguard.
Question 316
A team reports: A backend team needs Copilot to modernize a 2,000-line module but must keep public APIs unchanged. They observe: A single giant prompt rewrites the module and breaks several public methods. Which ONE technical decision most directly corrects the cause?
- Decompose the work into bounded functions and require public-signature and regression tests at each step.
- Identify the exact module, symbol and signature, and attach the relevant source excerpt.
- First request a bounded step-by-step diagnosis referencing the specific destructor sequence, then ask for a minimal patch.
- Identify the public API, allowed implementation region, compatibility requirement and diff boundaries explicitly.
Explanation
Correct Answer
A:The task exceeded a manageable verifiable change scope.
Incorrect Answers
B:Identifying the exact module, symbol and signature, and attaching the relevant source excerpt would not address the need for decompose the work into bounded functions and require public-signature and regression tests at each step.
C:The prompt jumped to a solution without requiring evidence of the underlying mechanism.
D:Identifying the public API, allowed implementation region, compatibility requirement and diff boundaries explicitly would not address the need for decompose the work into bounded functions and require public-signature and regression tests at each step.
Question 317
A team has implemented the following change: Decompose the work into bounded functions and require public-signature and regression tests at each step. The original requirement was: A backend team needs Copilot to modernize a 2,000-line module but must keep public APIs unchanged. Which ONE validation test best demonstrates the intended behavior?
- Use AddressSanitizer and a targeted lifetime test to verify the stated diagnosis.
- Refactor one helper, rerun API contract tests, then proceed to the next helper.
- Compare the answer to the selected method's body and call graph.
- Check imports from dependent modules before and after the refactor.
Explanation
Correct Answer
B:Refactor one helper, rerun API contract tests, then proceed to the next helper.
Incorrect Answers
A:Generated code requires independently defined acceptance tests and review against the real program contract. Verification of this result involves refactor one helper, rerun API contract tests, then proceed to the next helper.
C:Compare the answer to the selected method's body and call graph. That check cannot demonstrate compatibility with the pinned dependency version.
D:Approved dependency sources and lockfiles protect builds against invented or impersonated packages. Verification of this result involves refactor one helper, rerun API contract tests, then proceed to the next helper.
Question 318
During troubleshooting, investigators observe: A single giant prompt rewrites the module and breaks several public methods. The system context is: A backend team needs Copilot to modernize a 2,000-line module but must keep public APIs unchanged. Which ONE cause most directly explains the observed failure?
- The prompt jumped to a solution without requiring evidence of the underlying mechanism.
- An ambiguous referent let conversation history select the wrong target.
- The refactor prompt did not state protected interface constraints.
- The task exceeded a manageable verifiable change scope.
Explanation
Correct Answer
D:The remedy should therefore be: Decompose the work into bounded functions and require public-signature and regression tests at each step.
Incorrect Answers
A:The prompt jumped to a solution without requiring evidence of the underlying mechanism. The missing change involves the task exceeded a manageable verifiable change scope.
B:An ambiguous referent let conversation history select the wrong target. The task exceeded a manageable verifiable change scope.
C:The refactor prompt did not state protected interface constraints. The task exceeded a manageable verifiable change scope.
Question 319
The current design is: Decompose the work into bounded functions and require public-signature and regression tests at each step. The prior failure was: A single giant prompt rewrites the module and breaks several public methods. Which ONE ongoing control best protects against recurrence of that specific failure?
- Use explicit file and symbol references in follow-up questions.
- Use incremental decomposition and checkpoint tests for large AI-assisted edits.
- Preserve API signatures in tests and review any generated symbol renames.
- Use diagnosis-first prompting for complex defects and verify with tools.
Explanation
Correct Answer
B:The task exceeded a manageable verifiable change scope. Use incremental decomposition and checkpoint tests for large AI-assisted edits.
Incorrect Answers
A:Use explicit file and symbol references in follow-up questions. It would leave tested AI-assisted changes that preserve stated requirements without the specified safeguard.
C:Generated API calls must be verified against the installed library's actual supported signatures. It would leave tested AI-assisted changes that preserve stated requirements without the specified safeguard.
D:Use diagnosis-first prompting for complex defects and verify with tools. It would leave tested AI-assisted changes that preserve stated requirements without the specified safeguard.
Question 320
A team reports: Copilot generates Jest tests, but the repository uses Vitest and a custom fixture setup. They observe: The suggested test imports unsupported Jest globals. Which ONE technical decision most directly corrects the cause?
- Decompose the work into bounded functions and require public-signature and regression tests at each step.
- Identify the public API, allowed implementation region, compatibility requirement and diff boundaries explicitly.
- Iterate with measured feedback, concrete constraints and the failing profile instead of merely repeating "try again".
- Attach the existing package manifest and representative test file, and explicitly require Vitest conventions.
Explanation
Correct Answer
D:The model lacked versioned local test-framework context.
Incorrect Answers
A:Decompose the work into bounded functions and require public-signature and regression tests at each step would not attach the existing package manifest and representative test file, and explicitly require Vitest conventions.
B:Identifying the public API, allowed implementation region, compatibility requirement and diff boundaries explicitly would not attach the existing package manifest and representative test file, and explicitly require Vitest conventions.
C:Iterate with measured feedback, concrete constraints and the failing profile would not attach the existing package manifest and representative test file, and explicitly require Vitest conventions.
GitHub Copilot (GH-300) Practice Tests