←
Back to Dashboard
GitHub Copilot (GH-300)
Practice Exam
🧠 GH-300 Topic Mindmap
GITHUB COPILOT
├── FEATURES
│ ├── Code Completion (Ghost Text, FIM - Fill In Middle)
│ ├── Copilot Chat (/fix, /tests, /doc, /explain)
│ ├── Copilot CLI (Shell/Git help)
│ ├── Shortcuts (Tab=Accept, Alt+]=Next, Ctrl+Enter=Panel)
│ └── Voice ("Hey Code")
├── PLANS & PRIVACY
│ ├── Individual (Standard, trains on snippets unless disabled?)
│ ├── Business (Org management, seat assignment, NO training)
│ ├── Enterprise (Repo Indexing, Fine-Tuning, PR Summaries)
│ └── Filters (Public Code Matcher - Block or Allow)
├── PROMPT ENGINEERING
│ ├── Principles (Single, Specific, Short, Surround)
│ ├── Context (Open tabs, File cursor position, Imports)
│ ├── Iteration (Refining prompts based on output)
│ └── Zero/Few-shot (Examples vs No Examples)
├── SECURITY & GOVERNANCE
│ ├── Vulnerabilities (Hardcoded secrets, SQLi patterns)
│ ├── Licensing (Attribution for public code matches)
│ ├── Policies (Org level blocking of features)
│ └── HTTPS/Certificates (Proxy setup for corporate networks)
└── INTEGRATIONS
├── IDEs (VS Code, Visual Studio, JetBrains, Vim)
├── CLI (gh extension)
└── Mobile (Copilot in GitHub Mobile App)
1. Prompt Engineering (The "4 Ss")
Single: Give Copilot a single, clear task.
Specific: Be explicit about inputs, outputs, and format (e.g., "Return a JSON list").
Short: Keep prompts concise but descriptive.
Surround: Open content (tabs) provides context. Copilot reads neighboring files.
2. Key Features & Commands
Ghost Text: Gray text suggestions as you type. (Tab to accept).
Copilot Chat: /fix (debug), /tests (generate tests), /doc (document code), /explain
(explain code).
Copilot CLI: `gh copilot suggest "cmd"` or `gh copilot explain "cmd"`.
Voice: "Hey Code" to trigger voice commands (accessibility).
3. Privacy & Governance
Public Code Matcher: Setting to block suggestions that match public code (filters 150+
chars).
Data Usage: Copilot for Business/Enterprise does NOT retain snippets for training.
Context: Copilot uses the "Context Window" composed of the active file + open tabs to
generate relevant code.
Q1. You want Copilot to generate a unit test for the function you just highlighted. What
is the fastest way? (A) Type "Update this code" (B) Use the /tests slash command in Chat (C) Delete the
function (D) Open a new file
Answer: B - The `/tests` slash command in Copilot Chat is specifically designed to
scaffold unit tests for selected code.
Q2. Copilot is suggesting code that looks outdated. What is the best "Prompt
Engineering" fix? (A) Restart the IDE (B) Add a comment at the top specifying the library version (e.g.
"// Using React 18") (C) Turn off Copilot (D) Type faster
Answer: B - Providing context via comments (like library versions) guides Copilot to use
modern patterns.
Q3. Which feature allows you to ask "Where is the memory leak in this class?"? (A) Ghost
text (B) Copilot Chat (C) Copilot CLI (D) IntelliSense
Answer: B - Copilot Chat supports natural language Q&A about your codebase.
Q4. You are an admin. You want to ensure developers don't accidentally use copyrighted
code verbatim. (A) Disable Copilot (B) Enable "Block suggestions matching public code" policy (C) Use a
VPN (D) Delete the repo
Answer: B - The "Suggestions matching public code" filter blocks suggestions that match
~150 chars of public code on GitHub.
Q5. How does Copilot know about the other files in your project? (A) It scans your
entire hard drive (B) It reads all open tabs in your IDE (Neighboring Tabs) (C) It reads your mind (D)
It only knows the current file
Answer: B - Copilot uses "Neighboring Tabs" technique. Keeping relevant files open
improving suggestion quality.
Q6. You get a suggestion you like, but want to see alternatives. Shortcut? (A)
Ctrl+Enter (B) Alt+] (C) Ctrl+Z (D) Esc
Answer: A - Ctrl+Enter (or Cmd+Enter) opens the Copilot pane showing 10 synthesized
solutions at once.
Q7. Which slash command fixes a bug in the selected code? (A) /fix (B) /bug (C) /help
(D) /create
Answer: A - `/fix` analyzes the selected code (or error message) and proposes a
correction.
Q8. You are writing a Python script. You type `# Calculate average of list`. This
technique is: (A) Chatting (B) Comment-driven development (C) CLI (D) Debugging
Answer: B - Using comments to describe intent so Copilot generates the code block is
"Comment-driven development".
Q9. Copilot Enterprise allows indexing your org's code. What does this improve? (A)
Speed (B) Contextual relevance (RAG) for your specific internal libraries (C) Nothing (D) UI color
Answer: B - Repository indexing allows Copilot to "know" your proprietary APIs and
internal patterns better.
Q10. How do you document a legacy function using Copilot? (A) Rewrite it (B) Select
lines and use /doc (C) Use /fix (D) Use /tests
Answer: B - `/doc` generates documentation comments (JSDoc, Python docstrings) for the
selection.
Q11. You need a shell command to find all large files. (A) Copilot Chat (B) Copilot for
CLI (`gh copilot suggest`) (C) Copilot Word (D) Copilot Excel
Answer: B - GitHub Copilot in the CLI is specifically for shell commands, git commands,
and explanations.
Q12. Does Copilot for Business train on your codebase? (A) Yes, always (B) No (C) Only
on Fridays (D) Only Python files
Answer: B - Copilot for Business/Enterprise does not retain code snippets for training
the base model.
Q13. To improve prompt results, you should: (A) Be vague (B) Use examples (Few-shot) (C)
Close all tabs (D) Use binary
Answer: B - Providing examples (input -> output mapping) significantly improves the
model's accuracy.
Q14. What context is sent to Copilot? (A) Only the cursor line (B) The entire file +
open tabs (Context Window) (C) The git history (D) Your browser history
Answer: B - It sends the code before/after cursor and segments from other open files (up
to the token limit).
Q15. You want to translate code from Java to Python. (A) /fix (B) Ask Chat "Translate
this selection to Python" (C) Retype it manually (D) Use formatted strings
Answer: B - Copilot Chat is excellent for language translation tasks.
Q16. "Iterative Prompting" means: (A) Asking once and giving up (B) Rephrasing and
refining the prompt based on the previous output (C) Writing a loop (D) Rebooting
Answer: B - Refining the prompt conversationally to get closer to the desired result.
Q17. Which is a security best practice when using Copilot? (A) Accept all suggestions
blindly (B) Review all code for secrets and vulnerabilities (C) Turn off firewall (D) Share API keys
Answer: B - You are the pilot. Always review code. Copilot *can* hallucinate or suggest
insecure patterns.
Q18. Copilot didn't suggest anything. Why? (A) You are offline (B) File type is
unsupported (unlikely) (C) Content matched a filter (D) All of the above
Answer: D - Could be network, filter blocking (public code or profanity), or context
limits.
Q19. You want to generate a commit message. (A) Type `git commit` (B) Use the "Sparkle"
icon in Source Control view (C) /commit (D) Write it yourself
Answer: B - In VS Code, the sparkle/star icon in the Git view uses Copilot to generate a
message based on staged changes.
Q20. "Zero-shot" prompting means: (A) Providing no examples, just instruction (B)
Providing 0 instructions (C) Providing 100 examples (D) A failed prompt
Answer: A - "Zero-shot" is asking the model to do something without giving it a prior
example.
Q21. How to accept *part* of a suggestion (next word)? (A) Tab (B) Ctrl+RightArrow (C)
Enter (D) Space
Answer: B - Ctrl+RightArrow (or Cmd+Right) accepts the next word of the ghost text.
Q22. Can Copilot help with SQL? (A) No (B) Yes, it supports SQL queries (C) Only MySQL
(D) Only NoSQL
Answer: B - Copilot supports many languages including SQL.
Q23. Accessibility feature aimed at those who cannot type easily. (A) Copilot Voice /
Hey Code (B) Copilot Vision (C) Copilot Touch (D) Copilot Foot
Answer: A - Voice controls allow coding via speech.
Q24. Before using code generated by Copilot in production, you must: (A) Pray (B) Test,
Review, and Scan it (C) Deploy immediately (D) Encrypt it
Answer: B - AI code is not guaranteed to be correct or secure. Standard QC applies.
Q25. "Chain of Thought" prompting involves: (A) Asking the model to "Think step-by-step"
(B) Connecting computers (C) Using blockchain (D) Writing long lines
Answer: A - Prompting the model to explain its reasoning often improves accuracy on
complex logic.
Q26. You see a security vulnerability in Copilot code. (A) Ignore it (B) Fix it manually
or ask /fix "Make this secure" (C) Report it to FBI (D) Panic
Answer: B - You are responsible. Ask Copilot to remediate it.
Q27. Copilot creates a variable `const apiKey = "sk-12345"`. (A) Commit it (B) It's a
hallucinated or example secret, DO NOT use real keys (C) It's your real key (D) It's fine
Answer: B - Never hardcode secrets. Copilot might suggest dummy keys. Use ENV variables.
Q28. To scaffold an entire React component using "Boilerplate reduction": (A) Type
`import React` and class name (B) Type `// create a react component for...` (C) Both A and B (D) Copy
paste
Answer: C - Starting the structure or comment triggers Copilot's boilerplate generation.
Q29. Which extension is needed for VS Code? (A) GitHub Copilot (B) GitHub Actions (C)
Python (D) Docker
Answer: A - You need the GitHub Copilot extension (and optionally Copilot Chat).
Q30. /explain serves what purpose? (A) Runs the code (B) Explains what the selected code
does in plain English (C) Deletes code (D) Translates language
Answer: B - Useful for understanding legacy or complex regex/algorithms.
Q31. Copilot suggestions are "Stochastic". This means: (A) They are always correct (B)
They are probabilistic/random (C) They are licensed (D) They are slow
Answer: B - The model predicts the statistically likely next token. It does not "know"
truth.
Q32. Managing seat licenses is done where? (A) IDE (B) GitHub.com Organization Settings
(C) Azure Portal (D) AWS Console
Answer: B - Admins manage seats in GitHub Organization settings.
Q33. If you want Copilot to use a specific coding style: (A) Hope (B) Provide examples
of that style in the file/open tabs (C) Change settings (D) It's impossible
Answer: B - Copilot mimics the style found in the context (open files).
Q34. Can Copilot help with RegExp? (A) No (B) Yes, describe the pattern in a comment (C)
Only simple ones (D) Only numbers
Answer: B - It is exceptionally good at regex. e.g. `// Regex for verifying email
address`.
Q35. "Hallucination" in code looks like: (A) A red squiggle (B) Importing a library that
doesn't exist (C) A syntax error (D) A comment
Answer: B - AI typically hallucinates plausible-sounding but non-existent package names
or methods.
Q36. Best way to write a SQL query for a specific table schema? (A) Connect DB (B) Paste
the schema/CREATE TABLE statement in the file or a tab (C) Guess (D) Write SELECT *
Answer: B - Providing the schema/DDL in context helps Copilot generate correct column
names.
Q37. Can you use Copilot in Vim/Neovim? (A) No (B) Yes (C) Only Emacs (D) Only Nano
Answer: B - Copilot has plugins for Vim/Neovim, JetBrains, Visual Studio, and VS Code.
Q38. "Prompt injection" in this context is: (A) A security attack (B) Inserting code (C)
Not applicable to Code suggestion (D) Making prompts faster
Answer: C - Prompt Injection is more relevant to Chatbots, but malicious comments
*could* theoretically bias suggestions.
Q39. How to regenerate a suggestion? (A) Hover over the suggestion (B) Press Alt+]
(Cycle Next) (C) Close IDE (D) Press F5
Answer: B - Cycle through alternative suggestions using the keyboard shortcut (Alt+] or
Option+]).
Q40. Does Copilot run code? (A) Yes (B) No, it only suggests text (C) Yes, inside
sandbox (D) Yes, on GitHub servers
Answer: B - Copilot suggests text. It does not execute code or side effects.
Q41. A developer is stuck on a logic error. (A) Use /fix (B) Use /explain to understand
the logic (C) Ask Chat "What is wrong with this logic?" (D) All of the above
Answer: D - All are valid strategies.
Q42. Which language is Copilot best at? (A) Binary (B) Python/JS/TS/Go (C) Cobol (D)
English
Answer: B - It performs best on languages with large public training data (Python, JS,
TS, etc.), though it supports many.
Q43. "Comments are code". This philosophy means: (A) Comments are compiled (B) Comments
are the primary input for the AI to generate the code (C) Comments are useless (D) Comments are slow
Answer: B - In the AI era, comments serve as prompt instructions.
Q44. What happens if you lose internet? (A) Copilot stops working (B) It works offline
(C) It uses local cache (D) It switches to generic mode
Answer: A - Copilot requires an internet connection to reach the inference server.
Q45. To reduce "latency" of suggestions: (A) Close unused tabs (reduce context size) (B)
Buy faster internet (C) Type slower (D) All of the above
Answer: A - Reducing context *can* help fast processing, but usually it's network
dependent.
Q46. Copilot suggests an API that requires a paid key. (A) Copilot pays for it (B) You
must obtain the key yourself (C) It's free (D) Use it illegally
Answer: B - Copilot suggests code patterns; it does not provide access/subscriptions to
3rd party services.
Q47. Role of "System Prompt"? (A) User input (B) Hidden instructions from GitHub to the
model (C) Your code (D) The shell
Answer: B - The system prompt (hidden) guides the model's behavior (e.g. "You are a
helpful coding assistant").
Q48. The best way to use Copilot for a newly created file: (A) Start typing
imports/class names to establish context (B) Wait (C) Close it (D) Use binary
Answer: A - "Priming" the context by typing the basic structure helps Copilot infer what
you want to do.
Q49. You are an Organization Admin. You want to STOP Copilot from suggesting code that
matches public repositories. (A) It is impossible (B) Set "Suggestions matching public code" to BLOCKED
in Org Settings (C) Send an email to GitHub (D) Use a firewall
Answer: B - Organization settings allows Admins to BLOCK suggestions that match public
code (~150 chars) for all members.
Q50. Which feature is EXCLUSIVE to Copilot Enterprise? (A) Copilot Chat (B) Repository
Indexing / Knowledge Base (C) CLI (D) VS Code support
Answer: B - Indexing your organization's repositories for deeper context (RAG) is a
flagship Enterprise feature.
Q51. Your corporate proxy is blocking Copilot. Which domain needs to be whitelisted? (A)
google.com (B) githubcopilot.com (and related endpoints like `api.github.com`) (C) stackoverflow.com (D)
npmjs.org
Answer: B - You typically need to whitelist `https://api.github.com`,
`https://copilot-proxy.githubusercontent.com`, etc.
Q52. "Fill-In-The-Middle" (FIM) allows Copilot to: (A) Fill gaps *between* existing code
lines (B) Delete code (C) Only append to the end (D) Write comments only
Answer: A - Copilot looks at the prefix (before cursor) AND suffix (after cursor) to
fill in the missing middle code logically.
Q53. You want Copilot to help you write a Pull Request summary. (A) It can't (B) Use
Copilot Enterprise feature "Summarize PR" (C) Use /fix (D) Use CLI
Answer: B - Copilot Enterprise generates summaries for PRs (changes analysis)
automatically.
Q54. A developer pushes code with a hardcoded Secret Key suggested by Copilot. Who is at
fault? (A) GitHub (B) The Developer (C) The ISP (D) The IDE
Answer: B - The developer is always responsible for the code they commit. "AI suggested
it" is not a valid defense.
Q55. Copilot Chat is ignoring your specific requirement. How to fix? (A) Yell at it (B)
Provide "One-shot" or "Few-shot" examples in the chat prompt (C) Turn it off and on (D) Reinstall VS
Code
Answer: B - "Few-shot prompting" (giving 1-2 examples of input->output) is the most
effective way to correct style/logic.
Q56. Which IDE does NOT officially support Copilot? (A) VS Code (B) JetBrains IntelliJ
(C) Visual Studio (D) Arduino IDE 1.0
Answer: D - Copilot has extensions for VS Code, Visual Studio, JetBrains, Vim/Neovim,
and Azure Data Studio. Arduino IDE is not supported.
Q57. How often does the Copilot client update its "Context"? (A) Every keystroke/cursor
move (B) Once a day (C) On save (D) Never
Answer: A - Copilot continuously updates the prompt sent to the backend as you type and
move the cursor to ensure relevance.
Q58. Can Copilot write Infrastructure as Code (Terraform)? (A) No (B) Yes, it supports
HCL and Terraform (C) Only CloudFormation (D) Only Pulumi
Answer: B - Copilot is excellent at Terraform/HCL patterns if you provide the
provider/resource context.
Q59. You are using Copilot Individual. Can you prevent your snippets from being used for
training? (A) Yes, uncheck "Allow GitHub to use my code snippets for product improvements" (B) No, it's
mandatory (C) Only in Europe (D) Only if you pay more
Answer: A - Individual users can opt-out of data sharing (telemetry/snippets) in their
personal GitHub settings.
Q60. When using the CLI `gh copilot suggest`, can it execute the command for you? (A)
Yes, automatically (B) No, it only explains (C) It offers to "Run this command" after you confirm
(optional) (D) It runs it in a VM
Answer: C - For safety, it suggests the command and gives you an explicit option to
"Copy to clipboard" or "Execute command". It never auto-runs.
Q61. A user is added to an Organization with a "Copilot Business" seat. What happens to
their "Copilot Individual" sub? (A) Nothing (B) It should be cancelled to avoid double billing (C) It
converts to Enterprise (D) It becomes free
Answer: B - If your Org grants you a seat, you typically cancel your personal sub. The
Org seat overrides/takes precedence.
Q62. What is the maximum number of tabs Copilot can look at? (A) Infinite (B) Only 1 (C)
It's limited by the "Prompt Budget" / Token Limit (D) Exactly 5
Answer: C - Copilot intelligently selects the most relevant snippets from open tabs
until the prompt token limit is reached.
Q63. You want to exclude specific files (e.g. `secret.js`) from Copilot completely. (A)
Delete them (B) Use `.copilotignore` (Not quite) (C) VS Code Settings "Content Exclusion" (Copilot
Business/Enterprise feature) (D) Add a comment
Answer: C - Copilot Business/Enterprise allow Admins to specify "Content Exclusions" (by
path/repo) so Copilot ignores those files. Individual does not have this centrally.
Q64. "Prompt Crafting" best practice: (A) "Write code" (B) "Write a Python function to
parse CSV using pandas with error handling" (C) "Fix this" (D) "Do magic"
Answer: B - Be Specific. Mention the Language (Python), Task (Parse CSV), Library
(pandas), and Requirement (Error handling).
Q65. How does Copilot handle self-signed certificates in a corporate env? (A) It fails
(B) It ignores SSL (C) You must configure the `win-ca` or variable `NODE_EXTRA_CA_CERTS` (D) It uses VPN
Answer: C - Common corporate issue. You need to point the Copilot extension (Node.js
based) to the custom CA cert chain.
Q66. Which model powers Copilot Chat? (A) GPT-2 (B) A customized version of GPT-4 (C)
BERT (D) Llama 1
Answer: B - Copilot Chat utilizes advanced OpenAI models (like GPT-4) optimized for
coding tasks.
Q67. You highlight a block of code and press `Ctrl+I` (Inline Chat). You type "Make this
more readable". This is: (A) Refactoring (B) Deleting (C) Auditing (D) Compiling
Answer: A - Inline Chat is perfect for iterative Refactoring (renaming variables,
extracting methods) without leaving the editor flow.
Q68. Copilot suggested a library `fast-json-parser-2024` that doesn't exist. This is:
(A) A typo (B) Hallucination (Package Squatting risk) (C) A private package (D) A future package
Answer: B - Hallucinated package names are a security risk (supply chain attack if
someone registers that name). Verify every package.
Q69. In Copilot Chat, what does the `@workspace` agent do? (A) Nothing (B) Searches the
entire workspace context to answer questions (C) Deletes the workspace (D) Formats code
Answer: B - `@workspace` allows you to ask questions that require knowledge of the full
project structure, not just the open file.
Q70. Which shortcut allows you to cycle through the *previous* suggestion? (A) Alt+[ (B)
Alt+Z (C) Ctrl+Z (D) Esc
Answer: A - Alt+[ (Option+[ on Mac) goes back to the previous suggestion in the list.
Q71. You want to generate database migrations based on your schema.prisma file. (A) Open
schema.prisma, open a new migration file, wait/prompt (B) Close schema.prisma (C) Type "SQL" (D) Use
Paint
Answer: A - Opening the source file (schema.prisma) provides the necessary context for
Copilot to generate the corresponding migration SQL.
Q72. Can Copilot explain *why* a regex matches a string? (A) No (B) Yes, select it and
ask /explain (C) Only if it wrote it (D) Only in binary
Answer: B - Copilot is excellent at breaking down cryptic regex logic into plain English
explanations.
Q73. Your organization requires "Audit Logs" of all Copilot activity. Which plan is
required? (A) Individual (B) Business or Enterprise (C) Free (D) Team
Answer: B - Audit logs (who was assigned a seat, policy changes) are a feature of
Business/Enterprise plans. Individual has no audit logs.
Q74. What is the "Reference" log in Copilot? (A) A list of users (B) A feature that
notifies you if the suggested code matches public code (with a link to the repo) (C) An error log (D) A
chat history
Answer: B - If "Public Code Matcher" is configured to ALLOW, it can provide reference
logs linking the suggestion to the GitHub repo it resembles.
Q75. You want to write a Docstring for a function. You type `"""` and wait. This is an
example of: (A) Contextual triggering (B) A bug (C) A crash (D) Zero-shot
Answer: A - Typing the docstring delimiter signals intent, triggering Copilot to
generate the documentation based on the function body.
Q76. A developer wants to use Copilot to generate sensitive data (PII) for testing. (A)
Good idea (B) Bad idea. Use libraries like Faker instead (C) Copilot cannot generate data (D) It is
illegal
Answer: B - While Copilot *can* generate fake data, relying on it for PII schemas is
risky. Better to ask it to "use Faker library to generate X".
Q77. Does Copilot check your code for syntax errors *before* you accept? (A) Yes (B) No,
it suggests text. The IDE's linter checks syntax (C) Yes, it runs a compiler (D) It guarantees valid
code
Answer: B - Copilot suggests raw text. It might suggest code with missing braces. The
IDE linter (red squiggles) is what flags the syntax error.
Q78. To troubleshoot "Copilot not working", where do you check logs in VS Code? (A)
Output Panel -> "GitHub Copilot" (B) Terminal (C) Debug Console (D) Problems tab
Answer: A - The Output Panel dropdown has a specific channel for "GitHub Copilot"
showing connection/auth/suggestion logs.
Q79. Which language feature makes it HARDER for Copilot to suggest correct code? (A)
Strong Typing (B) Dynamic Typing with no type hints (C) Comments (D) Descriptive names
Answer: B - Lack of type hints and vague names reduces the context Copilot can use to
infer intent. Strong typing (TypeScript/Go) helps it.
Q80. "Seat Management" - If you remove a user from the Organization, what happens to
their Copilot access? (A) They keep it forever (B) They lose access immediately (C) They keep it for 30
days (D) Nothing
Answer: B - License assignment is tied to Org membership. Removing them revokes the
license immediately.
Q81. Can you use Copilot to write emails? (A) Only in VS Code text files (B) Yes, in
Outlook (C) No (D) Only in Gmail
Answer: A - If you are writing a markdown or text file in the IDE, yes. But it helps
with *code* primarily. (Microsoft 365 Copilot is for Outlook).
Q82. Which keyboard shortcut triggers the Inline Chat? (A) Ctrl+I / Cmd+I (B) Ctrl+C (C)
Ctrl+V (D) Alt+F4
Answer: A - Ctrl+I opens the inline chat dialog to instruct Copilot to edit the
highlighted code.
Q83. "System 1" vs "System 2" thinking in AI. Ghost text is: (A) System 2 (Slow,
deliberate) (B) System 1 (Fast, intuitive) (C) Neither (D) Both
Answer: B - Ghost text is "System 1" (Fast thinking). Chat is often "System 2"
(Deliberate, conversational).
Q84. You want to convert a CSS file to SCSS. (A) Rename file to .scss and wait (B) Open
both, ask Chat to "Convert style.css to SCSS format" (C) Retype (D) Use Paint
Answer: B - Chat is efficient at file-to-file transformation tasks.
Q85. Can Copilot help with `bash` scripts? (A) No (B) Yes (C) Only PowerShell (D) Only
Zsh
Answer: B - Copilot helps write shell scripts (bash, zsh, powershell) effectively.
Q86. What is the "Knowledge Base" in Copilot Enterprise? (A) A wikipedia link (B) A
collection of markdown documentation files you point Copilot to (C) A database (D) A secret
Answer: B - You can create curated Knowledge Bases (from repos of docs) to answer
questions specifically from *that* documentation.
Q87. Rate Limiting: If you request too many suggestions too fast? (A) You get banned (B)
Copilot may temporarily throttle requests (C) Your computer explodes (D) Nothing
Answer: B - To prevent abuse, there are rate limits. You may see a temporary pause in
suggestions.
Q88. How to troubleshoot if "Public Code Filter" is actually working? (A) Try to
generate a known unique block of Quake code (B) Check the logs (C) Trust it (D) Disable internet
Answer: B - Logs will indicate if a suggestion was "Blocked by public code filter
policy". (Trying to generate the "Invverse Square Root" code from Quake is a common test).
Q89. Can Copilot generate Mermaid.js diagrams? (A) No (B) Yes, describe the diagram in
text (C) Only SVG (D) Only PNG
Answer: B - Yes! Ask "Generate a mermaid flowchart for this logic" and it works great.
Q90. Which AI model parameter does Copilot optimize for code? (A) Creativity (B) Latency
and Correctness (C) Humor (D) Length
Answer: B - Coding requires low latency (speed) and high correctness (syntax).
Q91. Can you use Copilot to learn a NEW language? (A) No (B) Yes, it acts as a tutor
(e.g. "Explain what this Go code does for a Java dev") (C) Only Spanish (D) Only French
Answer: B - It is an excellent tutor. Cross-language explanation is a top use case.
Q92. The `github-copilot.enable` setting in VS Code allows you to: (A) Enable/Disable
globally or per-language (B) Change color (C) Change font (D) Uninstall
Answer: A - You can granularly enable/disable Copilot for specific languages (e.g.
disable for Markdown, enable for Python).
Q93. What is "Context Pruning"? (A) Deleting files (B) The AI engine selecting which
parts of open files to keep vs ignore to fit the token window (C) A bug (D) A clean up tool
Answer: B - Intelligent pruning ensures the most relevant code (imports, definitions)
stays in the prompt while dropping less relevant lines.
Q94. You want to ensure your Prompt is "Surrounded". What do you do? (A) Put the cursor
in the middle of the file (B) Write comments above and below (C) Open relevant files in other tabs (D)
All of the above
Answer: D - All help. "Surround" means providing context *around* the cursor (tabs,
previous lines, next lines).
Q95. Copilot suggestion is cut off. Why? (A) Token limit (B) It got tired (C) Screen
size (D) Font size
Answer: A - The output token limit (often not visible to user) will cut off long
generations. Use Chat to "Continue".
Q96. Can Copilot access your local terminal history? (A) Yes (B) No (unless you paste it
into chat) (C) Always (D) Only on Linux
Answer: B - By default, it does not scan terminal history. You must explicitly provide
that context (e.g. copy paste error into chat). @terminal is a new experimental agent.
Q97. What is the benefit of "Slash Commands" (/fix, /tests) over natural language? (A)
They look cool (B) They are optimized "System Prompts" tuned for that specific task (C) They are slower
(D) They cost more
Answer: B - Slash commands use highly-optimized, pre-engineered prompts under the hood
to give better results than generic queries.
Q98. To start a fresh context in Chat, you should: (A) Delete the chat history / New
Chat (B) Keep typing (C) Restart VS Code (D) Buy a new PC
Answer: A - Cleaning the chat history ensures the model isn't confused by previous,
unrelated turns.
Q99. Your org has strict IP rules. Which setting is critical? (A) Block Public Code
Matches (B) Enable Dark Mode (C) Enable Voice (D) Disable Python
Answer: A - Blocking public code matches protects the org from accidental GPL
contamination or copyright claims.
Q100. The ultimate rule of using Copilot: (A) Trust it completely (B) Trust but Verify
(C) Ignore it (D) Fear it
Answer: B - Trust but Verify. AI is a co-pilot, not the pilot. You are responsible for
the flight.