Q1.At a help desk, a primary Copilot Studio agent is the first to receive inquiries from users.
For certain requests, you want this primary agent to hand off processing to another existing Copilot Studio agent.
To meet the business requirements, you must enable the following behaviors for a connected agent.
– Allow it to be called by other agents.
– Make it available for other agents to use in collaboration.
– Allow other agents to call it during a conversation.
Which setting should you configure for each requirement? Assign the appropriate configuration to each requirement (each configuration may be used once, more than once, or not at all).
* Each correct answer is worth one point.
Drag each option to the matching requirement’s drop zone, or tap to place it. Click a filled zone to send it back. (The same option may be used more than once.)
Add a connected agent
Configure a connection to the connected agent
Add the connected agent as an indexed knowledge source
Show answer
To make a connected agent callable by other agents, you first publish the target agent so it becomes available as a connected agent.
An agent that has not been published cannot be referenced or selected by other agents.
Collaborative use and calling during a conversation are enabled by using “Add a connected agent” from the primary agent’s build screen to establish the connection.
Configuring a connection is for authenticating external APIs or tools, and an indexed knowledge source is for grounding on documents; neither is used for delegation between agents.
Add connected agents – Microsoft Copilot Studio
Q2.An electronics component manufacturer is building a Copilot Studio solution that answers policy questions, retrieves operational metrics from structured data, and performs authenticated updates to backend systems.
Rather than consolidating everything into a single large agent, the company has decided to combine multiple specialized agents.
Custom agents for each action are already available.
To meet the business requirements, the solution must do the following.
– Retrieve the latest metrics from structured data.
– Answer policy questions using curated knowledge.
– Perform authenticated updates during a conversation.
You must choose the appropriate agent design for each requirement.
Which approach should you use for each requirement? Assign the appropriate configuration to each requirement (each configuration may be used once, more than once, or not at all).
* Each correct answer is worth one point.
Drag each option to the matching requirement’s drop zone, or tap to place it. Click a filled zone to send it back. (The same option may be used more than once.)
Delegate the request to a tool-first agent
Delegate the request to a data-access agent
Delegate the request to a knowledge-focused agent
Show answer
In a multi-agent configuration, each requirement is delegated to the specialized agent best suited for it.
Retrieving the latest metrics from structured data is handled by a tool-first agent that runs APIs and connectors as tools.
Because authenticated update processing also centers on running actions, it is likewise delegated to the tool-first agent.
Answering policy questions with curated knowledge, on the other hand, is best suited to a knowledge-focused agent optimized for RAG over static documents.
A single agent or a data-access agent is not optimal for these requirements.
Multi-agent patterns – Microsoft Copilot Studio
Q3.An apparel retailer is extending its Copilot Studio agent by combining external capabilities with access to internal content.
Some user requests require grounded answers, while others require actions to be executed.
To meet the business requirements, the agent must do the following.
– Ground answers using search results from enterprise data.
– Call callable functions exposed through a standardized tool interface.
– Write updates to the system using authenticated actions during a conversation.
You must choose what to integrate to meet each requirement.
What should you integrate for each requirement? Assign the appropriate integration to each requirement (each integration may be used once, more than once, or not at all).
* Each correct answer is worth one point.
Drag each option to the matching requirement’s drop zone, or tap to place it. Click a filled zone to send it back. (The same option may be used more than once.)
Add a Dataverse table as a structured-data knowledge source
Add a Power Platform connector tool and call it from a topic
Add a Copilot connector knowledge source for indexed content
Show answer
Choose the appropriate integration for each requirement.
For grounding based on enterprise data, add a Copilot connector knowledge source that crawls content and ingests it into a search index.
To run callable functions through a standardized tool interface, connect to an MCP server and add its tools to the agent.
To write authenticated updates to the system during a conversation, add a Power Platform connector tool and call it from a topic.
Connectors support CRUD operations and credential mapping, making them well suited to authenticated actions.
Copilot connectors in Copilot Studio – Microsoft Copilot Studio
Q4.Note: This question is part of a series of questions that present the same scenario.
Each question includes a unique solution that might meet the stated goal, and there may be more than one correct solution or none at all.
After you answer this question, you cannot return to it to review it.
A company is configuring a Copilot Studio agent that uses generative answers nodes within multiple topics.
The responses must meet the following requirements.
– Use an executive-summary format.
– Apply only within specific topics.
– Continue to use the configured knowledge sources.
– Comply with the connector’s security controls.
You must configure a custom prompt to meet the formatting requirement.
Solution: Remove the generative answers node and rely solely on a custom prompt to generate the formatted response.
Does this solution meet the goal?
Show answer
Removing the generative answers node and relying only on a custom prompt does not meet the requirements.
This is because you must apply the executive-summary format only within specific topics while still preserving the configured knowledge sources and the connector’s security controls.
The correct approach is to override custom instructions in the properties of the generative answers node within the target topic, limiting the formatting at the node level.
Removing the node loses the knowledge references and security controls, so the goal cannot be achieved.
Therefore, the answer is “No.”
Configure generative answers nodes – Microsoft Copilot Studio
Q5.A shipping company’s Copilot Studio agent needs to retrieve real-time data from an external system that exposes a REST API.
This API returns structured JSON data and requires authentication.
To meet a user request, the agent must call this API in the middle of a conversation.
The solution must make the REST API call using a supported mechanism, and that call must be explicitly configured to run at runtime.
You must determine how to add the REST API to the agent.
Which solutions meet the goal? Choose two (each answer is a complete solution).
* Each correct answer is worth one point. (Multiple-select question)
Show answer
There are two ways to call an authenticated REST API at conversation runtime.
One is to configure an HTTP action (HTTP request node) within a topic and call it during the conversation, passing authentication in the headers and parsing the JSON.
The other is to add the REST API as a tool and call it from a topic or from the agent’s instructions.
By importing an OpenAPI specification, you can register it as a native REST tool.
Parsing variables with Formula, turning it into a knowledge source, relying on generative answers, or using a redirect node cannot achieve a runtime API call.
Extend an agent with actions using a REST API – Microsoft Copilot Studio
Q6.A development team is building an agent in Copilot Studio and wants to validate its behavior before broadening its use.
The team needs a way to test the agent repeatably and consistently using predefined conversations.
You must be able to evaluate the agent’s responses consistently across multiple test runs.
What should you do?
Show answer
To evaluate the agent’s behavior consistently before expanding access, define a test set consisting of prompts and expected results.
Using the Agent Evaluation feature in Copilot Studio, you can run automated regression tests repeatedly with predefined conversations.
This replaces subjective spot checks and lets you evaluate responses consistently across multiple test runs.
Reviewing scores from past sessions, monitoring telemetry, changing to qualitative evaluation, or publishing to additional channels are not appropriate ways to achieve reproducible, consistent evaluation.
Test and evaluate agents – Microsoft Copilot Studio
Q7.Note: This question is part of a series of questions that present the same scenario.
Each question includes a unique solution that might meet the stated goal, and there may be more than one correct solution or none at all.
After you answer this question, you cannot return to it to review it.
An agent uses a flow that calls an external service which occasionally fails or times out.
When a failure occurs, the agent must do the following.
– Not terminate silently.
– Send a notification that includes the error details.
You must configure the agent’s flow so that failures are handled in a controlled, predictable way.
Solution: Retry the failed action without performing any failure handling.
Does this solution meet the goal?
Show answer
Merely retrying the failed action without failure handling does not meet the requirements.
This is because, when the external service fails or times out, the flow must not terminate silently and must send a notification that includes the error details.
The correct approach is to configure “Run after” in Power Automate to create a failure path and place a notification action on it.
By default, the next step runs only if the previous step succeeded, so on failure the subsequent steps are skipped and the flow fails silently.
Retrying alone does not result in controlled, predictable failure handling, so the answer is “No.”
Error handling in Power Automate – Microsoft Power Automate
Q8.An insurance company is building a Copilot Studio agent that can answer questions using internal systems and also perform authenticated operations.
To meet the business requirements, the agent must do the following.
– Provide searchable answers from indexed internal content.
– Retrieve up-to-date facts at runtime without indexing the data.
– Perform authenticated update operations during a conversation.
You must choose the correct integration approach for each requirement.
Which option should you use for each requirement? Assign the appropriate solution to each requirement (each solution may be used once, more than once, or not at all).
* Each correct answer is worth one point.
Drag each option to the matching requirement’s drop zone, or tap to place it. Click a filled zone to send it back. (The same option may be used more than once.)
Copilot connector knowledge source
Power Platform connector for real-time knowledge
Show answer
Use a different integration method for each requirement.
To provide searchable answers from indexed internal content, use a Copilot connector knowledge source that ingests non-Microsoft data into Microsoft Graph.
To retrieve up-to-date facts at runtime without indexing the data, a Power Platform connector for real-time knowledge that connects directly to live line-of-business systems is appropriate.
Authenticated update operations during a conversation are performed with a Power Platform connector tool that can run actions and modify data.
The key to the decision is that the connector’s role differs between retrieval and updating (actions).
Overview of connectors for Microsoft 365 Copilot – Microsoft Learn
Q9.A company needs its Copilot Studio agent to be able to run operations exposed by an external MCP server.
This MCP server exposes callable tools and requires authentication.
The agent must be able to call the MCP tools in the middle of a conversation.
You must configure the MCP tools so the agent can run MCP tool calls.
What should you do?
Show answer
To run authenticated callable tools exposed by an external MCP server from the agent, configure authentication for the MCP server connection the agent uses.
Specifically, by setting up OAuth 2.0 authentication, the agent can securely run MCP tool calls.
Automatic calling via generative answers, replacing MCP with a real-time knowledge connector, or adding the MCP server output as an indexed knowledge source cannot run authenticated tools.
The reason this is the correct answer is that configuring authentication is required.
Add an existing MCP server to an agent – Microsoft Copilot Studio
Q10.A Copilot Studio agent has already been built in the development environment.
You must prepare this agent to be moved to the production environment using a supported, solution-based method.
You need to create an exportable package that includes the existing agent.
Which actions should you perform? Choose two (each answer is part of the solution).
* Each correct answer is worth one point. (Multiple-select question)
Show answer
To move a development-environment agent to production using a solution-based method, first create a custom solution and add the existing agent and the required objects.
Next, publish all customizations, convert the solution to a managed solution, and export it as an exportable package (.zip).
Deleting topics, publishing to a channel and exporting the channel configuration, or exporting while unmanaged are not the correct solution-based migration steps.
Therefore, the correct answers are creating a custom solution and converting it to managed.
Export and import agents – Microsoft Copilot Studio