GH-300 Official Certification Question Bank (125 Questions)
125 curated questions sourced directly from official Microsoft Learn study domains, GH-300 exam prep guides, and real test sets covering all core competency areas of the GH-300 exam.
Q81: What data source formats are supported when building Knowledge Bases in Copilot Enterprise?
Knowledge Bases in Copilot Enterprise index Markdown (.md) files stored within designated organization repositories.
- Indexes Markdown documentation files (.md).
- Enables grounded chat queries over internal docs.
Which file format is used to curate documentation for Copilot Enterprise Knowledge Bases?
Q82: What is the purpose of a Responsible AI Impact Assessment during product development?
An Impact Assessment evaluates potential societal risks, biases, safety hazards, and privacy concerns associated with deploying an AI solution before release.
- Evaluates safety, privacy, and bias risks prior to deployment.
- Standard compliance requirement under Microsoft's AI framework.
What is the goal of an AI Impact Assessment?
Q83: How can Copilot Chat assist developers during legacy software maintenance?
Developers can highlight legacy functions and use /explain or ask Copilot Chat to break down complex logic, identify legacy dependencies, and document undocumented code blocks.
- Accelerates legacy code onboarding and maintenance.
- Explains legacy regex, procedural math, and un-documented logic.
How does Copilot assist developers onboarding to legacy codebases?
Q84: What key immediately dismisses an inline ghost text suggestion without accepting it?
Pressing the Escape (Esc) key dismisses ghost text suggestions in the editor.
Esc: Dismisses inline suggestion.Tab: Accepts inline suggestion.
Which key dismisses an active ghost text completion suggestion?
Q85: Why are positive instructions ('Do X') generally more effective in Copilot prompts than negative constraints ('Don't do Y')?
LLMs can struggle to process complex negative constraints reliably. Stating explicit positive requirements ('Use Async/Await syntax') guides model attention better than negative prohibitions ('Don't use callbacks').
- Positive statements frame clear targets for model token processing.
- Pair negative rules with clear positive alternatives.
How should constraints be framed in prompts for consistent results?
Q86: What happens to billing when an admin removes a Copilot seat mid-month in Copilot Business?
Seat cancellations take effect at the end of the current billing cycle, or are pro-rated depending on subscription management settings.
- Seat allocations can be managed dynamically.
- Billing updates reflect active assigned user count.
Where do organization admins adjust seat allocations for Copilot Business?
Q87: What are 'Smart Actions' in VS Code Copilot Chat editor context menus?
Right-clicking highlighted code opens the Copilot Smart Actions context menu, offering direct shortcuts for common tasks like /explain, /fix, /tests, and /doc without typing commands manually.
- Right-click context menu shortcuts for key slash commands.
- Provides quick access to
/explain,/fix,/tests, and/doc.
How can a developer access Copilot actions without typing slash commands in the chat panel?
Q88: Which audit log event triggers when an admin changes the public code matching policy in an organization?
GitHub Organization Audit Logs capture policy updates under events such as copilot.cfb_exact_match or copilot.public_code_suggestions.
- Logs administrative setting changes for public code matching.
- Ensures administrative compliance tracking.
Where can compliance officers verify when Copilot policy settings were updated?
Q89: Where can custom instructions for Copilot be defined at a personal user level in VS Code?
Personal custom instructions can be added to VS Code User Settings under github.copilot.chat.codeGeneration.instructions, applying instructions across all personal chat sessions.
- Personal instructions: VS Code User Settings.
- Project instructions:
.github/copilot-instructions.md.
Where can a developer set personal default guidelines for all their local Copilot Chat sessions?
Q90: What defines a 'Zero-Shot Prompt' when querying Copilot?
A Zero-Shot Prompt asks the model to perform a task directly without providing prior examples or sample input/output demonstrations.
- Zero-Shot: Direct request with no examples.
- Few-Shot: Provides sample input/output demonstrations.
A prompt that asks Copilot to write a function without including any sample code examples is known as what?