GH-700 · Sample questions

GH-700 sample questions with answers and explanations

10 real questions from the CertifyDeck GH-700 bank, spread across the domains of the official GitHub Certified: Agentic AI Business Professional skills outline, with the correct answer and a full explanation of why the other options are wrong. They are original questions written from the public syllabus, not exam dumps, and they are the same style you will find in the practice tests and mock exams.

1. Question 1Domain: Enhance productivity by using GitHub CopilotMedium

   You are a business analyst at a clothing brand who uses Copilot Chat on the GitHub website. You finished an exploratory conversation and you want that conversation removed, not merely started over. Which slash command should you use?

   • A./rename
   • B./clear
   • C./delete
   • D./new
   Show answer and explanation

   Correct answer: C. /delete

   Explanation

   On the GitHub website, /delete deletes a conversation, which matches the need to remove it. /clear is the tempting choice, but the documented purpose is to clear the conversation, and /new only starts another one, so neither is the command for deleting that conversation.

   Official references

   • GitHub Copilot Chat cheat sheet
2. Question 2Domain: Enhance productivity by using GitHub CopilotMedium

   During an internal audit, a reviewer asks why a commit authored by Copilot changed an approval rule in a process file. The reviewer did not start the task and needs to see the agent's reasoning and the tools it used. What should the reviewer do?

   • A.Run /chronicle search for the file in Copilot CLI
   • B.Read the co-author listed in the commit message for the task
   • C.Check that the commit shows as Verified on GitHub
   • D.Follow the link in the commit message to the session logs
   Show answer and explanation

   Correct answer: D. Follow the link in the commit message to the session logs

   Explanation

   Each agent-authored commit message includes a link to the session logs, which show Copilot's internal reasoning and the tools it used, so a reviewer can trace why a change was made. The /chronicle command is tempting, but it searches your own session history, and this reviewer did not start the task. The Verified status confirms the commit's signature, and the co-author only identifies who started the work.

   Official references

   • Managing agent sessions
   • Using GitHub Copilot CLI session data
3. Question 3Domain: Enhance productivity by using GitHub CopilotEasy

   You are a compliance lead defining a custom agent that reviews policy documents in a repository. It must be able to read and search files, but it must not have the ability to change them. Which configuration should you use in the agent profile?

   • A.Set the tools property to an empty list
   • B.Set the tools property to all tools with a star
   • C.Omit the tools property and describe the role
   • D.Set the tools property to read and search
   Show answer and explanation

   Correct answer: D. Set the tools property to read and search

   Explanation

   Providing a list of specific aliases enables only those tools, so read and search give the agent what it needs without any editing ability. Omitting the property or using a star enables all available tools, including editing. An empty list is tempting as the strictest choice, but it disables all tools, so the agent could not read or search either.

   Official references

   • Custom agents configuration
   • Comparing Copilot customization features
4. Question 4Domain: Design and implement agentic workflows and automationsMedium

   An agent at a utility company should read new service requests and post a comment on each one. Your security lead says the agent job must not hold direct write access. How should you configure the workflow?

   • A.Use permissions: read-all, with no safe-outputs section at all
   • B.Use permissions: {} and ask for comments in the instructions
   • C.Grant issues: write in permissions, with no safe-outputs section
   • D.Use read-only permissions with the add-comment safe output
   Show answer and explanation

   Correct answer: D. Use read-only permissions with the add-comment safe output

   Explanation

   Workflows use read-only permissions by default, and the add-comment safe output lets the agent request a comment that a separate, permission-controlled job then posts. Granting issues: write would allow the comment but gives direct write access, which is exactly what the security lead ruled out.

   Official references

   • GitHub Tools Read Permissions
   • Safe Outputs
5. Question 5Domain: Design and implement agentic workflows and automationsEasy

   A manager at a shipping firm reviews a workflow scheduled as weekly on monday and asks why it does not name an exact cron time instead. The report only has to arrive on Monday. What is the documented benefit of this schedule?

   • A.It guarantees that the workflow starts at the first minute of the working day
   • B.It scatters run times to avoid load spikes and keeps each time consistent
   • C.It runs only when there has been activity in the repository during the week
   • D.It picks a different random time on every run so that no two runs ever match
   Show answer and explanation

   Correct answer: B. It scatters run times to avoid load spikes and keeps each time consistent

   Explanation

   Fuzzy schedules such as weekly on monday distribute execution times to prevent load spikes, and each workflow gets a deterministic time that stays the same across recompiles. A different random time on every run is the tempting reading, but the scattering is based on the workflow file path, so the time does not change between runs.

   Official references

   • Schedule Syntax
6. Question 6Domain: Perform human-in-the-loop tasks for agentic workflowsMedium

   A workflow file has safe-outputs: steer: true, with issues: read in its permissions. While a long run is in progress, a stakeholder wants to give the agent a correction. What should the stakeholder do?

   • A.Comment on the run's issue, including the keyword steer
   • B.Edit the Markdown instructions and compile the workflow again
   • C.Add an @copilot comment to the compiled lock file
   • D.Cancel the run and dispatch it again with the correction
   Show answer and explanation

   Correct answer: A. Comment on the run's issue, including the keyword steer

   Explanation

   With steer set to true, a run-scoped issue is created before the agent starts, and users add comments containing the keyword steer that the agent is told to read. Cancelling and dispatching again would also deliver the correction, but it discards the run in progress rather than steering it.

   Official references

   • Safe Outputs
7. Question 7Domain: Perform human-in-the-loop tasks for agentic workflowsEasy

   A program manager has four tasks for a Copilot session and is considering Autopilot mode to save time. Which task is the best fit for that mode?

   • A.Fix a reproducible test failure and verify the test suite
   • B.Explore several possible directions for a new customer portal
   • C.Decide which vendor requirements matter most to the business
   • D.Brainstorm how the team could improve its release process
   Show answer and explanation

   Correct answer: A. Fix a reproducible test failure and verify the test suite

   Explanation

   Autopilot mode suits tasks with a clear outcome and enough detail for the agent to decide how to implement them, such as fixing a reproducible failure. Brainstorming and exploring are open-ended, so they are more likely to need your guidance and may not meet your expectations.

   Official references

   • Allowing GitHub Copilot to work autonomously
8. Question 8Domain: Use GitHub platform featuresEasy

   You are a program manager at an insurer. Compliance wants Copilot Chat on github.com to direct every member of your organization to the internal compliance knowledge base for regulatory questions, in any repository. Which type of customization should you recommend?

   • A.Organization custom instructions
   • B.A prompt file in one repository
   • C.Path-specific custom instructions
   • D.Personal instructions for each member
   Show answer and explanation

   Correct answer: A. Organization custom instructions

   Explanation

   Organization custom instructions provide organization-wide language, security, compliance or support guidance and can direct people to an internal knowledge base. A prompt file is the tempting alternative, but people choose when to use it, so it would not apply to every member automatically.

   Official references

   • Custom instructions for GitHub Copilot
9. Question 9Domain: Use GitHub platform featuresMedium

   A technical project manager at a car rental company wants Copilot's agent to search a repository, update some documentation files and run the existing tests. A colleague proposes adding a new MCP server first. What should you recommend?

   • A.Add the new MCP server, since agents cannot edit files without one
   • B.Package the MCP server in a plugin before the agent starts the work
   • C.Rely on the agent's existing tools and skip the new server
   • D.Replace the repository instructions with the new MCP server's tools
   Show answer and explanation

   Correct answer: C. Rely on the agent's existing tools and skip the new server

   Explanation

   GitHub says to avoid adding an MCP server when the built-in tools already cover your needs, and lists searching a repository, changing files and running tests as work for an agent's existing tools. The claim that agents need a server to edit files is false, because editing is a built-in ability. MCP also adds connections to tools and data rather than replacing guidance such as instructions.

   Official references

   • Comparing Copilot customization features
10. Question 10Domain: Use GitHub platform featuresMedium

    Your team made changes on a branch named policy-update. They must end up in the branch named release-plan, not in the default branch. How should you set up the pull request?

    • A.Base branch policy-update and compare branch release-plan
    • B.Base branch release-plan and compare branch policy-update
    • C.Base branch set to the default branch and compare branch release-plan
    • D.Base branch and compare branch both set to release-plan
    Show answer and explanation

    Correct answer: B. Base branch release-plan and compare branch policy-update

    Explanation

    The base branch is where the changes will be merged and the compare branch is the topic branch that holds them, so release-plan is the base and policy-update is the compare branch. Swapping them is the tempting mistake, but it would propose merging release-plan into policy-update, the opposite of the requirement.

    Official references

    • Creating a pull request