100% Latest & Updated Microsoft AB-620 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 290
In an inventory analyst, the agent calls a stock API but occasionally treats a timeout as a zero stock count. The team selected this design: Distinguish unavailable data from valid numeric zero in tool output handling. Which implementation action BEST puts it into practice?
- Return a typed unavailable/error state on timeout instead of coercing to zero.
- Use supported token renewal and return an authentication-required error when needed.
- Configure the server connection, authentication, tool descriptions and supported operations.
- Add the connector operation as a tool and map its required parameters.
Explanation
Correct Answer
A:Return a typed unavailable/error state on timeout instead of coercing to zero. The expected result is that a simulated timeout is reported as unavailable, not sold out.
Incorrect Answers
B:An OAuth-backed tool's token expires while an agent is processing a multi-step request. In that setting, “Use supported token renewal and return an authentication-required error when needed” is relevant; the required choice here is “Return a typed unavailable/error state on timeout instead of coercing to zero”.
C:A Copilot Studio agent must invoke a supported order lookup exposed through a published MCP server. That condition could justify “Configure the server connection, authentication, tool descriptions and supported operations”. Here, “Return a typed unavailable/error state on timeout instead of coercing to zero” targets the needed outcome.
D:“Add the connector operation as a tool and map its required parameters” could be appropriate when an existing custom connector already defines order status operations and authentication. The present task instead requires “Return a typed unavailable/error state on timeout instead of coercing to zero”.
Question 291
In an inventory analyst, the agent calls a stock API but occasionally treats a timeout as a zero stock count. After the team implements this action: Return a typed unavailable/error state on timeout instead of coercing to zero. Which evidence BEST verifies the requirement?
- A test conversation invokes the expected MCP tool with correctly shaped arguments.
- Repeated interaction attempts do not create multiple service orders.
- Order number input invokes the existing connector and returns a typed status.
- A simulated timeout is reported as unavailable, not sold out.
Explanation
Correct Answer
D:A simulated timeout is reported as unavailable, not sold out. This result tests the effect of return a typed unavailable/error state on timeout instead of coercing to zero.
Incorrect Answers
A:A Copilot Studio agent must invoke a supported order lookup exposed through a published MCP server. In that setting, “A test conversation invokes the expected MCP tool with correctly shaped arguments” is relevant; the required choice here is “A simulated timeout is reported as unavailable, not sold out”.
B:Computer use executes an action repeatedly when a page is slow to confirm the result. That condition could justify “Repeated interaction attempts do not create multiple service orders”. Here, “A simulated timeout is reported as unavailable, not sold out” targets the needed outcome.
C:“Order number input invokes the existing connector and returns a typed status” could be appropriate when an existing custom connector already defines order status operations and authentication. The present task instead requires “A simulated timeout is reported as unavailable, not sold out”.
Question 292
In an enterprise architect, two API tools have near-identical descriptions, and the agent selects the wrong record-updater. Which solution design decision BEST addresses this requirement?
- Distinguish unavailable data from valid numeric zero in tool output handling.
- Separate privileged deletion actions from ordinary retrieval capability.
- Improve tool metadata and restrict each tool to a distinct operation.
- Branch on HTTP status and tool-result success metadata.
Explanation
Correct Answer
C:Improve tool metadata and restrict each tool to a distinct operation. Implement this design by give each tool a clear action-oriented description, schema and permissions.
Incorrect Answers
A:The agent calls a stock API but occasionally treats a timeout as a zero stock count. That condition could justify “Distinguish unavailable data from valid numeric zero in tool output handling”. Here, “Improve tool metadata and restrict each tool to a distinct operation” targets the needed outcome.
B:“Separate privileged deletion actions from ordinary retrieval capability” could be appropriate when a custom API tool can delete records, although the agent only needs read access. The present task instead requires “Improve tool metadata and restrict each tool to a distinct operation”.
D:A REST endpoint returns an error object, yet the agent prints it as a success response. That condition could justify “Branch on HTTP status and tool-result success metadata”. Here, “Improve tool metadata and restrict each tool to a distinct operation” targets the needed outcome.
Question 293
In an enterprise architect, two API tools have near-identical descriptions, and the agent selects the wrong record-updater. The team selected this design: Improve tool metadata and restrict each tool to a distinct operation. Which implementation action BEST puts it into practice?
- Give each tool a clear action-oriented description, schema and permissions.
- Constrain navigation and tool actions to the approved user task and resist page-borne instructions.
- Add the connector operation as a tool and map its required parameters.
- Send only required fields through an authenticated endpoint and redact unneeded attributes.
Explanation
Correct Answer
A:Give each tool a clear action-oriented description, schema and permissions. The expected result is that agent traces show read-only requests never call the update endpoint.
Incorrect Answers
B:A computer-use tool opens an untrusted webpage containing instructions to exfiltrate data. In that setting, “Constrain navigation and tool actions to the approved user task and resist page-borne instructions” is relevant; the required choice here is “Give each tool a clear action-oriented description, schema and permissions”.
C:An existing custom connector already defines order status operations and authentication. That condition could justify “Add the connector operation as a tool and map its required parameters”. Here, “Give each tool a clear action-oriented description, schema and permissions” targets the needed outcome.
D:“Send only required fields through an authenticated endpoint and redact unneeded attributes” could be appropriate when a tool sends email addresses to an external REST endpoint but lacks purpose-limited field handling. The present task instead requires “Give each tool a clear action-oriented description, schema and permissions”.
Question 294
In an enterprise architect, two API tools have near-identical descriptions, and the agent selects the wrong record-updater. After the team implements this action: Give each tool a clear action-oriented description, schema and permissions. Which evidence BEST verifies the requirement?
- Agent traces show read-only requests never call the update endpoint.
- Burst tests complete without an uncontrolled retry storm.
- A denied request leaves the form in draft state.
- Conversation logic works when a provider returns missing optional fields.
Explanation
Correct Answer
A:Agent traces show read-only requests never call the update endpoint. This result tests the effect of give each tool a clear action-oriented description, schema and permissions.
Incorrect Answers
B:An API responds with HTTP 429 when too many agent calls occur concurrently. That condition could justify “Burst tests complete without an uncontrolled retry storm”. Here, “Agent traces show read-only requests never call the update endpoint” targets the needed outcome.
C:“A denied request leaves the form in draft state” could be appropriate when a computer-use tool would click a final 'Submit' button that creates a legal record. The present task instead requires “Agent traces show read-only requests never call the update endpoint”.
D:An agent needs to call an existing weathered internal API with a nonstandard response shape. In that setting, “Conversation logic works when a provider returns missing optional fields” is relevant; the required choice here is “Agent traces show read-only requests never call the update endpoint”.
Question 295
In an audit office, an MCP server exposes tool names but provides insufficient descriptions of side effects. Which solution design decision BEST addresses this requirement?
- Add a trusted MCP server tool rather than copying its endpoint into agent instructions.
- Review the tool catalog and require explicit side-effect contracts before use.
- Improve tool metadata and restrict each tool to a distinct operation.
- Treat webpage text as untrusted task data rather than new agent authority.
Explanation
Correct Answer
B:Review the tool catalog and require explicit side-effect contracts before use. Implement this design by clarify capabilities, required inputs and mutation risk before connecting the tools to production.
Incorrect Answers
A:A Copilot Studio agent must invoke a supported order lookup exposed through a published MCP server. That condition could justify “Add a trusted MCP server tool rather than copying its endpoint into agent instructions”. Here, “Review the tool catalog and require explicit side-effect contracts before use” targets the needed outcome.
C:Two API tools have near-identical descriptions, and the agent selects the wrong record-updater. In that setting, “Improve tool metadata and restrict each tool to a distinct operation” is relevant; the required choice here is “Review the tool catalog and require explicit side-effect contracts before use”.
D:A computer-use tool opens an untrusted webpage containing instructions to exfiltrate data. That condition could justify “Treat webpage text as untrusted task data rather than new agent authority”. Here, “Review the tool catalog and require explicit side-effect contracts before use” targets the needed outcome.
Question 296
In an audit office, an MCP server exposes tool names but provides insufficient descriptions of side effects. The team selected this design: Review the tool catalog and require explicit side-effect contracts before use. Which implementation action BEST puts it into practice?
- Bind the connector to a valid connection in the destination environment and retest.
- Constrain navigation and tool actions to the approved user task and resist page-borne instructions.
- Clarify capabilities, required inputs and mutation risk before connecting the tools to production.
- Use supported token renewal and return an authentication-required error when needed.
Explanation
Correct Answer
C:Clarify capabilities, required inputs and mutation risk before connecting the tools to production. The expected result is that high-risk tool calls are not selected under ambiguous descriptions.
Incorrect Answers
A:“Bind the connector to a valid connection in the destination environment and retest” could be appropriate when a custom connector fails only after moving an agent from development to test. The present task instead requires “Clarify capabilities, required inputs and mutation risk before connecting the tools to production”.
B:A computer-use tool opens an untrusted webpage containing instructions to exfiltrate data. In that setting, “Constrain navigation and tool actions to the approved user task and resist page-borne instructions” is relevant; the required choice here is “Clarify capabilities, required inputs and mutation risk before connecting the tools to production”.
D:“Use supported token renewal and return an authentication-required error when needed” could be appropriate when an OAuth-backed tool's token expires while an agent is processing a multi-step request. The present task instead requires “Clarify capabilities, required inputs and mutation risk before connecting the tools to production”.
Question 297
In an audit office, an MCP server exposes tool names but provides insufficient descriptions of side effects. After the team implements this action: Clarify capabilities, required inputs and mutation risk before connecting the tools to production. Which evidence BEST verifies the requirement?
- The tool succeeds under test-environment credentials without using development secrets.
- High-risk tool calls are not selected under ambiguous descriptions.
- Requests rejected for missing fields no longer reach the MCP server.
- Burst tests complete without an uncontrolled retry storm.
Explanation
Correct Answer
B:High-risk tool calls are not selected under ambiguous descriptions. This result tests the effect of clarify capabilities, required inputs and mutation risk before connecting the tools to production.
Incorrect Answers
A:A custom connector fails only after moving an agent from development to test. In that setting, “The tool succeeds under test-environment credentials without using development secrets” is relevant; the required choice here is “High-risk tool calls are not selected under ambiguous descriptions”.
C:“Requests rejected for missing fields no longer reach the MCP server” could be appropriate when an MCP service can invoke an update function but requires a field absent from the agent's tool schema. The present task instead requires “High-risk tool calls are not selected under ambiguous descriptions”.
D:An API responds with HTTP 429 when too many agent calls occur concurrently. In that setting, “Burst tests complete without an uncontrolled retry storm” is relevant; the required choice here is “High-risk tool calls are not selected under ambiguous descriptions”.
Question 298
In a utility company, computer use executes an action repeatedly when a page is slow to confirm the result. Which solution design decision BEST addresses this requirement?
- Normalize the provider-specific response into a stable tool output contract.
- Explicitly register the workflow as a tool rather than instruct the agent to hallucinate it.
- Add a transaction-state check and approval-safe retry design for computer-use actions.
- Add completion checks and prevent duplicate transactions across retries.
Explanation
Correct Answer
D:Add completion checks and prevent duplicate transactions across retries. Implement this design by verifying page outcome or transaction ID before resubmitting the action.
Incorrect Answers
A:An agent needs to call an existing weathered internal API with a nonstandard response shape. That condition could justify “Normalize the provider-specific response into a stable tool output contract”. Here, “Add completion checks and prevent duplicate transactions across retries” targets the needed outcome.
B:“Explicitly register the workflow as a tool rather than instruct the agent to hallucinate it” could be appropriate when a user asks the agent to run a workflow that is not attached as an available tool. The present task instead requires “Add completion checks and prevent duplicate transactions across retries”.
C:A web UI automation can unintentionally submit the same payment twice after a network drop. In that setting, “Add a transaction-state check and approval-safe retry design for computer-use actions” is relevant; the required choice here is “Add completion checks and prevent duplicate transactions across retries”.
Question 299
In a utility company, computer use executes an action repeatedly when a page is slow to confirm the result. The team selected this design: Add completion checks and prevent duplicate transactions across retries. Which implementation action BEST puts it into practice?
- Inspect the MCP handshake, invoke a safe test call and correct metadata mismatches.
- Verify page outcome or transaction ID before resubmitting the action.
- Require an authorized user to approve before submitting the legal form.
- Use continuation tokens until the approved result boundary is reached.
Explanation
Correct Answer
B:Verify page outcome or transaction ID before resubmitting the action. The expected result is that repeated interaction attempts do not create multiple service orders.
Incorrect Answers
A:“Inspect the MCP handshake, invoke a safe test call and correct metadata mismatches” could be appropriate when an MCP endpoint is reachable but exposes a tool whose description does not match the actual result. The present task instead requires “Verify page outcome or transaction ID before resubmitting the action”.
C:A computer-use tool would click a final 'Submit' button that creates a legal record. That condition could justify “Require an authorized user to approve before submitting the legal form”. Here, “Verify page outcome or transaction ID before resubmitting the action” targets the needed outcome.
D:“Use continuation tokens until the approved result boundary is reached” could be appropriate when a REST API uses pagination and the agent answers from only the first page of records. The present task instead requires “Verify page outcome or transaction ID before resubmitting the action”.
Question 300
In a utility company, computer use executes an action repeatedly when a page is slow to confirm the result. After the team implements this action: Verify page outcome or transaction ID before resubmitting the action. Which evidence BEST verifies the requirement?
- A prompt-injection test cannot make the agent visit an unauthorized destination.
- Repeated interaction attempts do not create multiple service orders.
- A user request to delete data cannot be satisfied using the read-only tool.
- A denied request leaves the form in draft state.
Explanation
Correct Answer
B:Repeated interaction attempts do not create multiple service orders. This result tests the effect of verifying page outcome or transaction ID before resubmitting the action.
Incorrect Answers
A:A computer-use tool opens an untrusted webpage containing instructions to exfiltrate data. In that setting, “A prompt-injection test cannot make the agent visit an unauthorized destination” is relevant; the required choice here is “Repeated interaction attempts do not create multiple service orders”.
C:“A user request to delete data cannot be satisfied using the read-only tool” could be appropriate when a custom API tool can delete records, although the agent only needs read access. The present task instead requires “Repeated interaction attempts do not create multiple service orders”.
D:A computer-use tool would click a final 'Submit' button that creates a legal record. In that setting, “A denied request leaves the form in draft state” is relevant; the required choice here is “Repeated interaction attempts do not create multiple service orders”.
Designing and Building Integrated AI Agent Solutions in Copilot Studio (AB-620) Practice Tests