Independent: not affiliated with GitHub. Free to try; the study bundle adds a study guide and two practice exams.
5 original questions, none of them in the practice exams.
Pick an answer. Show answer explains every option and links its GitHub source.
5 questions / Explanations link GitHub sources / Last updated Oct 26, 2026
Question 1
Multiple choice
Domain 1 / Task 1.2 Validate and operate AI tools
Copilot code review comments on a pull request for a billing service that the refund handler can issue the same refund twice when two requests arrive together, and it offers a suggested change that wraps the handler in a lock. The pull request's tests pass, but none of them sends two requests at once. What should the author do next?
Show answerHide answer
Answer: C
Why
GitHub's application card for Copilot agents lists two limitations of Copilot code review that apply here. It may highlight problems that do not exist or that rest on a misunderstanding of the code, so its comments should be carefully reviewed and considered before taking action. Its code suggestions may appear valid without being semantically or syntactically correct, or without resolving the problem named in the comment, so code from Copilot should always be reviewed and tested. The passing tests never send two requests at once, so they settle neither the claim nor the fix.
Why not the others
AThe card says generated code may appear valid without being correct or without resolving the problem named in the comment, and that code from Copilot should always be reviewed and tested.
BThe card says model switching is not supported for Copilot code review, which uses a carefully tuned mix of models, prompts and system behaviors.
DThe card says comments should be carefully reviewed and considered before taking action, and tests that never send two requests at once cannot show that a duplicate refund is impossible.
An organization owner on Copilot Business wants Copilot Chat on GitHub.com to apply the same secure-coding conventions to every member's requests, across all of the organization's repositories, without adding a file to each repository. Where should the conventions go?
Show answerHide answer
Answer: D
Why
GitHub's custom instructions page says organization owners can add instructions to tailor Copilot to the needs and preferences of their organization, that these instructions are used for all members of the organization, and that they can only be configured for organizations with a Copilot Business or Copilot Enterprise subscription. They are currently supported for Copilot Chat, Copilot code review and Copilot cloud agent on GitHub.com, so they cover Chat on GitHub.com without a file in any repository.
Why not the others
APersonal instructions apply to the person who set them, not to other members.
BA prompt file is used only when a member chooses it in a chat, so the conventions would not apply to every request.
CRepository-wide instructions apply to requests made in the context of that one repository, not across the organization.
A security team's rule says that no model provider may keep data from a Copilot request after the response has been returned. An organization owner on Copilot Business wants to enable the Gemini models for Copilot Chat and asks the security team to approve. What should the team conclude from GitHub's documentation on model hosting?
Show answerHide answer
Answer: A
Why
GitHub's model hosting page says Copilot uses Gemini models hosted on Google Cloud Platform and sends prompts and metadata to GCP, which commits that Gemini does not use the prompts or its responses to train its models. The same section says GitHub uses Google's Durable Caching: cached data is encrypted and held for up to 24 hours, and it is never used to train or fine-tune models. Data held for up to 24 hours is data kept after the response, so the rule is not met, even though the data is never used for training.
Why not the others
BThe page describes a zero data retention agreement for the OpenAI models; its Gemini section describes cached data held for up to 24 hours instead.
CThat commitment covers training only; the same section says Google's Durable Caching holds cached data for up to 24 hours.
DThe page says cached data is never used to train or fine-tune models, and that Gemini does not use prompts or responses to train its models.
A developer's Copilot Chat prompt for a CSV import function opens with several paragraphs about the team's history and naming habits, puts the one hard requirement (reject any row without a customer ID) in the middle, and ends with a request for logging. The responses keep leaving out the customer-ID check. How should the developer restructure the prompt?
Show answerHide answer
Answer: B
Why
GitHub's prompt engineering guidance says to give Copilot a broad description of the goal first and then list the specific requirements, with the most important requirements first, because Copilot gives extra weight to instructions at the beginning and end of a prompt. The rule in the middle of the current prompt is in neither position.
Why not the others
AA general pointer to standards replaces a specific requirement with an ambiguous one; the guidance is to list specific requirements and avoid ambiguity.
CThe guidance is to start general, with the goal, and then get specific; this order puts the details first and the goal last.
DA label leaves the rule in the middle of the prompt; the guidance is to put the most important requirements first.
Domain 6 / Task 6.2 Safeguards and troubleshooting
A repository administrator on Copilot Business excluded the repository's internal/sdk/ directory from Copilot. Two reports follow. In VS Code, inline suggestions in a file outside internal/sdk/ filled in the correct fields of a type that is defined only in internal/sdk/. In Xcode, Copilot Chat answered a question by quoting a function from an excluded file. Which two documented limits of content exclusion explain the reports? Choose two.
Show answerHide answer
Answers: A and D
Why
GitHub's content exclusion page documents both limits. Copilot may use semantic information from an excluded file when the IDE provides it indirectly, such as type information and hover-over definitions for symbols used in code, which explains the inline suggestions in VS Code. Its support table shows Xcode honoring content exclusion for inline suggestions but not for chat, which explains the quoted function.
Why not the others
BThe public preview note covers the GitHub website and GitHub Mobile, not IDEs.
CA repository administrator's rule affects any Copilot user in the enterprise working in that repository, in supported editors as well as in code review.
EIn VS Code, Visual Studio and JetBrains IDEs, Copilot Chat supports content exclusion, and excluded content does not inform Copilot's responses there.
The study bundle has two practice exams with 120 original questions, an answer key and an explanation for every answer, plus the 58-page study guide. Practice exam explanations are shorter than the ones on this page: a short paragraph per question, with no source links.
Study bundle
$19
120 practice questions / Study guide / Answer keys
Every bundle includes 6 months of access: download the PDFs, and any new editions, until 6 months after your purchase date. Your receipt has a license key. Enter it at overqual.com/download for the three PDFs, stamped with your email and order number. They are yours to keep.