CertPractice

GH-300 study guide

GH-300: GitHub Copilot practice questions

Responsible AI, GitHub Copilot in the IDE and CLI, agent mode and MCP, code review, organization policies, data flow, prompt engineering, testing, and privacy safeguards. Mapped to the skills measured as of August 7, 2026.

Start free practice Official from Microsoft: Exam page · Study guide (skills measured)

Skills measured

6 skill areas and 41 objectives, as of August 7, 2026. Weights show the share of the exam each area makes up.

Use GitHub Copilot responsibly (15-20%)
  • 1.1 Describe risks and limitations of Generative AI tools
  • 1.2 Describe ethical and responsible AI usage
  • 1.3 Identify potential harms and mitigation strategies of AI usage
  • 1.4 Explain the need to validate AI output
  • 1.5 Identify how to operate GitHub Copilot responsibly
Use GitHub Copilot features (25-30%)
  • 2.1 Enable Copilot in the IDE
  • 2.2 Trigger Copilot through inline suggestions, chat, CLI, and agent mode
  • 2.3 Configure content exclusions for specific files or repositories
  • 2.4 Define GitHub Copilot CLI and how it benefits developers
  • 2.5 Identify the steps for installing GitHub Copilot CLI
  • 2.6 Describe key GitHub Copilot CLI features and commands
  • 2.7 Use GitHub Copilot CLI interactively and in sessions
  • 2.8 Generate scripts and manage files with GitHub Copilot CLI
  • 2.9 Use Agent Mode, Copilot Edits, and MCP; manage Agent Sessions and delegate tasks to Sub-Agents
  • 2.10 Use Copilot for code review and coding assistance
  • 2.11 Utilize Spaces, Spark, Pull Request summaries, and custom review standards via instructions files
  • 2.12 Understand the limits, options, feedback, and commands of GitHub Copilot Chat, including prompt files
  • 2.13 Configure organization-wide policies, Copilot code review policies and feature availability across IDEs and github.com
  • 2.14 Utilize audit log events
  • 2.15 Manage subscriptions using the REST API
Understand GitHub Copilot data and architecture (10-15%)
  • 3.1 Explain data usage, flow, and sharing
  • 3.2 Describe input processing and prompt building
  • 3.3 Explain proxy filtering and post-processing
  • 3.4 Visualize code suggestion lifecycle
  • 3.5 Describe limitations of LLMs and Copilot
Apply prompt engineering and context crafting (10-15%)
  • 4.1 Describe prompt structure and context
  • 4.2 Understand how context is determined
  • 4.3 Use zero-shot and few-shot prompting
  • 4.4 Apply best practices for prompt crafting
  • 4.5 Explain prompt engineering principles
  • 4.6 Describe prompt process flow and chat history usage
Improve developer productivity with GitHub Copilot (10-15%)
  • 5.1 Use Copilot for code generation, refactoring, and documentation
  • 5.2 Accelerate learning and reduce context switching
  • 5.3 Generate sample data and modernize legacy code
  • 5.4 Generate unit and integration tests
  • 5.5 Identify edge cases and write assertions
  • 5.6 Suggest security improvements and performance optimizations
Configure privacy, content exclusions, and safeguards (10-15%)
  • 6.1 Configure content exclusions and editor settings
  • 6.2 Describe ownership and limitations of outputs
  • 6.3 Enable suggestions matching public code filtering
  • 6.4 Resolve issues with suggestions and content exclusions

Sample questions

Sample question 1 · Use GitHub Copilot responsibly

Copilot Chat recommends calling a method named parseStrict() on a popular JSON library, but the method does not exist in any version of that library. Which limitation of generative AI does this illustrate?

  1. A. Rate limiting applied by the Copilot service
  2. B. Public code filtering removing the correct answer
  3. C. Content exclusion hiding the real method name
  4. D. Hallucination of plausible but nonexistent APIs
Show answer and explanation

Answer: D

Large language models predict likely text, so they can invent APIs that look real but do not exist. Rate limits, content exclusions and the public code filter do not fabricate method names.

Learn more: Application card: GitHub Copilot Chat

Sample question 2 · Use GitHub Copilot features

A developer was just assigned a Copilot Business seat and wants Copilot suggestions in Visual Studio Code. What must the developer do?

  1. A. Ask a repository admin to turn on Copilot in the settings of each repository to be used
  2. B. Paste a Copilot API key copied from the organization settings into the VS Code user settings file
  3. C. Install the GitHub Copilot extension and sign in with the GitHub account that has the seat
  4. D. Install the GitHub CLI and run a login command that links VS Code to the organization
Show answer and explanation

Answer: C

In VS Code, Copilot is enabled by installing the GitHub Copilot extension and signing in with a GitHub account that has Copilot access. There is no API key to paste, and seats are not enabled per repository.

Learn more: Installing the GitHub Copilot extension in your environment · Configuring GitHub Copilot in your environment

Sample question 3 · Understand GitHub Copilot data and architecture

A security reviewer asks how code snippets travel from a developer's IDE to the GitHub Copilot service. What is the correct answer?

  1. A. They are sent unencrypted because the service runs inside GitHub
  2. B. They are encrypted in transit using TLS
  3. C. They are stored in the repository and synced later
  4. D. They are emailed to the service in batches
Show answer and explanation

Answer: B

Prompts and suggestions are transmitted over encrypted TLS connections between the IDE and the Copilot service. Copilot does not batch, email, or commit prompt data to the repository.

Learn more: GitHub Terms of Service · Hosting of models for GitHub Copilot

Sample question 4 · Apply prompt engineering and context crafting

Which three elements make up a strong prompt for Copilot Chat?

  1. A. A long friendly greeting, a joke, and a request for a fast answer
  2. B. The developer's name, the current date, and the installed IDE version
  3. C. A clear goal, relevant context, and expected output or constraints
  4. D. A list of keywords without any full sentences
Show answer and explanation

Answer: C

Effective prompts state what you want, give the context needed, and describe constraints or the output format. Greetings, metadata, and bare keyword lists add little useful guidance.

Learn more: Prompt engineering for GitHub Copilot

Sample question 5 · Improve developer productivity with GitHub Copilot

In Visual Studio Code, a developer selects a Python function and wants Copilot Chat to add a docstring describing its parameters and return value. Which slash command fits this task?

  1. A. /explain
  2. B. /doc
  3. C. /new
  4. D. /fix
Show answer and explanation

Answer: B

The /doc command asks Copilot to generate documentation comments, such as docstrings, for the selected code. /explain describes code in chat without adding comments, /fix proposes corrections for problems, and /new scaffolds a new project.

Learn more: Improving code readability and maintainability · Create README

Frequently asked questions

How many GH-300 practice questions are there?

218 original questions covering all 41 objectives in Microsoft's skills outline (skills measured as of August 7, 2026). Each one has an explanation, and most link to the relevant Microsoft Learn page.

Is it free?

The first 10 questions are free, with no account needed. Unlocking the full set costs US$9.99 for 30 days of access.

Are these real exam questions?

No. Every question is written from scratch and mapped to the published skills outline. They are not copied from the real exam, so they help you learn the material rather than memorize answers.

How does practice differ from exam mode?

Practice mode lets you check or view the answer after each question. Exam mode is timed (100 minutes), shuffles questions and answers, and scores you at the end.

Practice all GH-300 questions