Microsoft Updated GH-600 Exam Questions and Answers by jaden

Statement

Answer

The Run Review handoff resolves to review.agent.md.

Yes

The Start Implementation handoff fails to identify the target agent.

No

Selecting Start Implementation switches the chat to the implementer agent and automatically submits the handoff prompt to that agent by using GPT-5.2 (copilot).

Yes

A handoff identifies the destination agent separately from the label displayed to the user. The Run Review entry targets code-review, which is the declared name of the agent defined in review.agent.md. The filename therefore does not need to match the handoff’s visible button label or the declared agent name.

The Start Implementation entry identifies the implementation agent. The uppercase display name IMPLEMENTER does not, in this scenario, make the existing implementation handoff invalid. Its purpose remains to transfer the planning conversation into the implementation role.

The send property determines what happens to the handoff prompt after the transition. With send: true, selecting the handoff submits the prompt automatically instead of merely placing it in the input box for editing. The model property selects the model for that handoff using the qualified model-and-provider format. Successful use assumes that the specified model is available to the user.

These properties control destination, prompt submission, and model selection independently; the button label is only the user-facing action name.

Study-guide topics: agent routing, handoff configuration, and execution transitions. Reference: VS Code—Custom agent handoffs and frontmatter .

===============

Statement

Selection

The command will prevent runaway loops.

Yes

The command will allow the agent to use all tools.

Yes

The command will allow targeted human intervention at key decision points.

No

The first statement is Yes in the context of autopilot continuation. The option --max-autopilot-continues 10 places a ceiling on automatic continuation messages, preventing the agent from initiating an unlimited sequence of those continuations. This limit should not be confused with a timeout for every subprocess or a guarantee that an individual command cannot hang.

The second statement is Yes because --yolo is an alias for granting all permissions. It authorizes tool, path, and URL access without ordinary permission prompts, subject to any stronger enforced restrictions. The script therefore provides broad authority rather than a restricted dependency-maintenance permission set.

The third statement is No. Combining programmatic prompting with autopilot and broad permissions is designed for execution without routine human interaction. The command does not establish explicit approval checkpoints before consequential actions. An operator’s ability to interrupt execution is different from a workflow that deliberately pauses for review.

The Bash options improve shell error handling but do not create agent approval gates or replace the continuation limit.

Relevant curriculum topics are autonomy levels, execution bounds, and the distinction between permission approval and autonomous continuation.

Workflow artifacts provide the required combination of retained execution evidence and file transfer between jobs. The agent-run job uploads trace.json and plan.md, associating both files with the workflow run. Reviewers can access the resulting artifact through the run’s GitHub Actions interface, while downstream jobs download it to their own execution environments.

This is necessary because separate jobs do not automatically share a local filesystem. Declaring a dependency establishes execution order but does not transfer the generated files. The review and deployment stages therefore need explicit artifact downloads in addition to their appropriate job dependencies.

Committing transient traces and plans introduces repository changes solely to transport run-specific evidence. A network share adds external infrastructure and does not directly satisfy the requirement for standard Actions artifact access. Dependency caching is intended for reusable inputs such as downloaded dependencies; it is not the appropriate mechanism for authoritative outputs from a particular agent run.

Artifacts preserve the plan and trace that informed the review, allowing subsequent deployment activity to use the corresponding files. Relevant curriculum topics are inspectable agent artifacts, observability, and human intervention.

Requirement

Control

Agent’s draft pull request workflow control

Approve and run the workflows from the pull request merge box.

Production deployment job control

In the deployment job, configure the required reviewers for the production environment.

Approving and running workflows from the pull request interface controls whether workflows associated with the agent’s draft pull request are allowed to begin. This provides the required human authorization before CI executes in that context.

Required reviewers on the production environment protect the deployment job itself. When the workflow reaches that environment, GitHub pauses the job until an authorized reviewer explicitly approves it. The approval is therefore applied at the production boundary, independently from the earlier pull-request workflow authorization.

A concurrency group controls overlap but does not require human approval. A custom deployment protection rule may be useful for specialized external validation, but it is not required by the stated scenario. Commit signing and run-summary inspection do not control the requested execution gates.

Study-guide topics: approval gates, protected environments, draft pull-request workflows, and production accountability.

===============