Practice Exams:

Microsoft AI-103 Dumps

EXAM: Developing AI Apps and Agents on Azure
Verified By Experts
Premium Bundle
$49.99

AI-103 Premium ETE File


  • Premium File 187 Questions & Answers
Whats Included:
  • Latest Questions
  • 100% Accurate Answers
  • Fast Exam Updates

Microsoft AI-103 Practice Test 46

Question 450

During a support escalation, a compliance specialist must accept a change for tool result untrusted content. The issue to be addressed is: A web-search result instructs an agent to leak its configured credentials. Which concrete verification action demonstrates that the relevant behavior works?

  1. Return a malicious test result and verify no unauthorized action
  2. Execute any instruction quoted in a retrieved web page as higher-level policy
  3. Simulate rejection and verify purchase does not execute
  4. Treat tool output as lower-trust evidence and ignore embedded action commands

Explanation

Correct Answer

A:

To verify this change, the payment services team should return a malicious test result and verify no unauthorized action. Operators still need the implementation to treat tool output as lower-trust evidence and ignore embedded action commands and the preventive requirement to enforce action allowlists and adversarial tool-output tests.

Incorrect Answers

B:

In the payment services team acceptance review, trying to execute any instruction quoted in a retrieved web page as higher-level policy does not test the exact tool result untrusted content condition where a web-search result instructs an agent to leak its configured credentials; the relevant test is to return a malicious test result and verify no unauthorized action. Acceptance should instead demonstrate that the service can return a malicious test result and verify no unauthorized action, using the described failure as the test criterion.

C:

For sequential approval workflow, it can be sensible to simulate rejection and verify purchase does not execute if a procurement agent must validate a request before seeking human purchase approval. Without the actual change to return a malicious test result and verify no unauthorized action, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

The action to treat tool output as lower-trust evidence and ignore embedded action commands can contribute to tool result untrusted content as a implementation step; here, a web-search result instructs an agent to leak its configured credentials. In the payment services team acceptance review, trying to treat tool output as lower-trust evidence and ignore embedded action commands does not test the exact tool result untrusted content condition where a web-search result instructs an agent to leak its configured credentials; the relevant test is to return a malicious test result and verify no unauthorized action.

Question 451

At a manufacturing group, the product support lead is preparing a staged service rollout. The concrete requirement is: A remote inventory API times out while an agent is making a shipping decision. Which design action BEST addresses the stated need?

  1. Retry forever while presenting the unfinished action as successful
  2. Record correlated traces for model calls tool invocations and handoffs
  3. Set tool timeouts and idempotent retry policies
  4. Simulate delayed responses and inspect user-facing error handling
  5. Use bounded retries or a safe fallback with explicit uncertain status

Explanation

Correct Answer

E:

The proposed implementation is to use bounded retries or a safe fallback with explicit uncertain status; this matches the operational requirement that a remote inventory API times out while an agent is making a shipping decision. The team can simulate delayed responses and inspect user-facing error handling. The team should also set tool timeouts and idempotent retry policies.

Incorrect Answers

A:

Choosing to retry forever while presenting the unfinished action as successful cannot directly replace the specified action to use bounded retries or a safe fallback with explicit uncertain status when a remote inventory API times out while an agent is making a shipping decision. The actual implementation must use bounded retries or a safe fallback with explicit uncertain status, because the stated service condition is that a remote inventory API times out while an agent is making a shipping decision.

B:

The option to record correlated traces for model calls tool invocations and handoffs would normally serve a different agent tracing across tools requirement, where a complex agent fails after several model calls but operators lack the failure step. The alternative may be useful in its own context, but it does not directly supply use bounded retries or a safe fallback with explicit uncertain status for the stated decision.

C:

The action to set tool timeouts and idempotent retry policies can contribute to tool timeout recovery as a ongoing preventive rule step; here, a remote inventory API times out while an agent is making a shipping decision. Choosing to set tool timeouts and idempotent retry policies cannot directly replace the specified action to use bounded retries or a safe fallback with explicit uncertain status when a remote inventory API times out while an agent is making a shipping decision.

D:

The action to simulate delayed responses and inspect user-facing error handling can contribute to tool timeout recovery as a targeted acceptance check step; here, a remote inventory API times out while an agent is making a shipping decision. Choosing to simulate delayed responses and inspect user-facing error handling cannot directly replace the specified action to use bounded retries or a safe fallback with explicit uncertain status when a remote inventory API times out while an agent is making a shipping decision.

Question 452

After a release acceptance review at a legal operations unit, the team observes this failure: A remote inventory API times out while an agent is making a shipping decision. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Return unstructured prose and assume downstream automation will parse it correctly
  2. Use bounded retries or a safe fallback with explicit uncertain status
  3. Retry forever while presenting the unfinished action as successful
  4. Simulate delayed responses and inspect user-facing error handling

Explanation

Correct Answer

C:

This design mistake is to retry forever while presenting the unfinished action as successful; it conflicts with the actual tool timeout recovery behavior needed when a remote inventory API times out while an agent is making a shipping decision. The legal operations unit should instead use bounded retries or a safe fallback with explicit uncertain status, followed by a check to simulate delayed responses and inspect user-facing error handling.

Incorrect Answers

A:

This legal operations unit failure-analysis alternative of return unstructured prose and assume downstream automation will parse it correctly does not diagnose the particular tool timeout recovery defect where a remote inventory API times out while an agent is making a shipping decision; the keyed mistake is retry forever while presenting the unfinished action as successful. The alternative may be useful in its own context, but it does not directly supply retry forever while presenting the unfinished action as successful for the stated decision.

B:

The alternative to use bounded retries or a safe fallback with explicit uncertain status is a implementation step rather than the faulty assumption being diagnosed. This legal operations unit failure-analysis alternative of use bounded retries or a safe fallback with explicit uncertain status does not diagnose the particular tool timeout recovery defect where a remote inventory API times out while an agent is making a shipping decision; the keyed mistake is retry forever while presenting the unfinished action as successful.

D:

The alternative to simulate delayed responses and inspect user-facing error handling is a targeted acceptance check step rather than the faulty assumption being diagnosed. This legal operations unit failure-analysis alternative of simulate delayed responses and inspect user-facing error handling does not diagnose the particular tool timeout recovery defect where a remote inventory API times out while an agent is making a shipping decision; the keyed mistake is retry forever while presenting the unfinished action as successful.

Question 453

The financial services division is responding to a design meeting. A service administrator discovers this issue: A remote inventory API times out while an agent is making a shipping decision. What corrective step should the team take to restore the intended functionality most directly?

  1. Retry forever while presenting the unfinished action as successful
  2. Record the timeout and present a recoverable operation state
  3. Simulate delayed responses and inspect user-facing error handling
  4. Repair the search connector and reindex approved documents

Explanation

Correct Answer

B:

The direct repair is to record the timeout and present a recoverable operation state. Once the condition is corrected, the financial services division should simulate delayed responses and inspect user-facing error handling. To reduce recurrence, it should also set tool timeouts and idempotent retry policies; those steps provide different recovery and prevention functions.

Incorrect Answers

A:

During financial services division remediation, the proposed step to retry forever while presenting the unfinished action as successful is not the direct restoration action for tool timeout recovery when a remote inventory API times out while an agent is making a shipping decision; the keyed repair is record the timeout and present a recoverable operation state. After remediation the team must simulate delayed responses and inspect user-facing error handling; the action in this option does not itself establish that target recovery outcome.

C:

The action to simulate delayed responses and inspect user-facing error handling can contribute to tool timeout recovery as a targeted acceptance check step; here, a remote inventory API times out while an agent is making a shipping decision. During financial services division remediation, the proposed step to simulate delayed responses and inspect user-facing error handling is not the direct restoration action for tool timeout recovery when a remote inventory API times out while an agent is making a shipping decision; the keyed repair is record the timeout and present a recoverable operation state.

D:

This approach relates to search tool integration: an agent needs current internal documentation to answer product questions. If the team chose it instead, the targeted check to simulate delayed responses and inspect user-facing error handling could still fail because the underlying required action is to record the timeout and present a recoverable operation state.

Question 454

A cloud developer at a logistics operator wants the following issue to be prevented in later deployments: A remote inventory API times out while an agent is making a shipping decision. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Record the timeout and present a recoverable operation state
  2. Scope search permissions to the user and validate citation coverage
  3. Set tool timeouts and idempotent retry policies
  4. Retry forever while presenting the unfinished action as successful

Explanation

Correct Answer

C:

Standardizing the boundary to set tool timeouts and idempotent retry policies is preventive because the same tool timeout recovery failure could otherwise return: a remote inventory API times out while an agent is making a shipping decision. For this logistics operator, the service should also use bounded retries or a safe fallback with explicit uncertain status, and tests should simulate delayed responses and inspect user-facing error handling so that the ongoing requirement is measurable.

Incorrect Answers

A:

The action to record the timeout and present a recoverable operation state can contribute to tool timeout recovery as a corrective remediation step; here, a remote inventory API times out while an agent is making a shipping decision. For prevention at the logistics operator, the measure to record the timeout and present a recoverable operation state is weaker or aimed at another issue than the tool timeout recovery guardrail to set tool timeouts and idempotent retry policies needed when a remote inventory API times out while an agent is making a shipping decision.

B:

The action to scope search permissions to the user and validate citation coverage addresses search tool integration rather than the decisive tool timeout recovery condition. The logistics operator should not infer that fixing one adjacent AI service validates its unrelated workflow; the requested action is to set tool timeouts and idempotent retry policies.

D:

For prevention at the logistics operator, the measure to retry forever while presenting the unfinished action as successful is weaker or aimed at another issue than the tool timeout recovery guardrail to set tool timeouts and idempotent retry policies needed when a remote inventory API times out while an agent is making a shipping decision. The lasting guardrail for this specific threat is to set tool timeouts and idempotent retry policies, not simply to complete an adjacent service task.

Question 455

During a release acceptance review, a site reliability engineer must accept a change for tool timeout recovery. The issue to be addressed is: A remote inventory API times out while an agent is making a shipping decision. Which concrete verification action demonstrates that the relevant behavior works?

  1. Use bounded retries or a safe fallback with explicit uncertain status
  2. Resume a long workflow and test that a prohibited action remains blocked
  3. Simulate delayed responses and inspect user-facing error handling
  4. Retry forever while presenting the unfinished action as successful

Explanation

Correct Answer

C:

The targeted acceptance test is to simulate delayed responses and inspect user-facing error handling. The design should first use bounded retries or a safe fallback with explicit uncertain status. A robust energy cooperative deployment also needs to set tool timeouts and idempotent retry policies; passing this check provides relevant evidence but does not certify every possible input or security risk.

Incorrect Answers

A:

The action to use bounded retries or a safe fallback with explicit uncertain status can contribute to tool timeout recovery as a implementation step; here, a remote inventory API times out while an agent is making a shipping decision. In the energy cooperative acceptance review, trying to use bounded retries or a safe fallback with explicit uncertain status does not test the exact tool timeout recovery condition where a remote inventory API times out while an agent is making a shipping decision; the relevant test is to simulate delayed responses and inspect user-facing error handling.

B:

For agent context compaction, it can be sensible to resume a long workflow and test that a prohibited action remains blocked if a long-running agent exhausts its context and forgets constraints. Without the actual change to simulate delayed responses and inspect user-facing error handling, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

D:

In the energy cooperative acceptance review, trying to retry forever while presenting the unfinished action as successful does not test the exact tool timeout recovery condition where a remote inventory API times out while an agent is making a shipping decision; the relevant test is to simulate delayed responses and inspect user-facing error handling. Acceptance should instead demonstrate that the service can simulate delayed responses and inspect user-facing error handling, using the described failure as the test criterion.

Question 456

At a payment services team, the security engineer is preparing a quality regression test. The concrete requirement is: A payment agent retries after a network disconnect and creates a duplicate purchase. Which design action BEST addresses the stated need?

  1. Use idempotency keys for retryable side-effecting tool operations
  2. Issue a fresh purchase request on every transient error without deduplication
  3. Treat tool output as lower-trust evidence and ignore embedded action commands
  4. Simulate a dropped response and confirm exactly one committed purchase

Explanation

Correct Answer

A:

Choosing to use idempotency keys for retryable side-effecting tool operations directly targets the described idempotent agent actions gap: a payment agent retries after a network disconnect and creates a duplicate purchase. As a lasting boundary, they also need to require durable operation IDs and duplicate-transaction checks, because an implementation by itself does not prove safe operation.

Incorrect Answers

B:

Choosing to issue a fresh purchase request on every transient error without deduplication cannot directly replace the specified action to use idempotency keys for retryable side-effecting tool operations when a payment agent retries after a network disconnect and creates a duplicate purchase. The actual implementation must use idempotency keys for retryable side-effecting tool operations, because the stated service condition is that a payment agent retries after a network disconnect and creates a duplicate purchase.

C:

Choosing to treat tool output as lower-trust evidence and ignore embedded action commands would direct effort toward tool result untrusted content, whose characteristic issue is that a web-search result instructs an agent to leak its configured credentials. The relevant corrective or preventive action here remains to use idempotency keys for retryable side-effecting tool operations, with its own validation criteria.

D:

The action to simulate a dropped response and confirm exactly one committed purchase can contribute to idempotent agent actions as a targeted acceptance check step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. Choosing to simulate a dropped response and confirm exactly one committed purchase cannot directly replace the specified action to use idempotency keys for retryable side-effecting tool operations when a payment agent retries after a network disconnect and creates a duplicate purchase.

Question 457

After a production incident at a transport authority, the team observes this failure: A payment agent retries after a network disconnect and creates a duplicate purchase. Which flawed assumption or implementation choice would MOST DIRECTLY explain why the intended outcome was not achieved?

  1. Assume a broad instruction to help users safely limits tool behavior
  2. Issue a fresh purchase request on every transient error without deduplication
  3. Use idempotency keys for retryable side-effecting tool operations
  4. Simulate a dropped response and confirm exactly one committed purchase

Explanation

Correct Answer

B:

The faulty assumption is issue a fresh purchase request on every transient error without deduplication. A functioning alternative would be to use idempotency keys for retryable side-effecting tool operations, and the relevant validation would be to simulate a dropped response and confirm exactly one committed purchase.

Incorrect Answers

A:

This transport authority failure-analysis alternative of assume a broad instruction to help users safely limits tool behavior does not diagnose the particular idempotent agent actions defect where a payment agent retries after a network disconnect and creates a duplicate purchase; the keyed mistake is issue a fresh purchase request on every transient error without deduplication. The alternative may be useful in its own context, but it does not directly supply issue a fresh purchase request on every transient error without deduplication for the stated decision.

C:

The alternative to use idempotency keys for retryable side-effecting tool operations is a implementation step rather than the faulty assumption being diagnosed. This transport authority failure-analysis alternative of use idempotency keys for retryable side-effecting tool operations does not diagnose the particular idempotent agent actions defect where a payment agent retries after a network disconnect and creates a duplicate purchase; the keyed mistake is issue a fresh purchase request on every transient error without deduplication.

D:

The alternative to simulate a dropped response and confirm exactly one committed purchase is a targeted acceptance check step rather than the faulty assumption being diagnosed. This transport authority failure-analysis alternative of simulate a dropped response and confirm exactly one committed purchase does not diagnose the particular idempotent agent actions defect where a payment agent retries after a network disconnect and creates a duplicate purchase; the keyed mistake is issue a fresh purchase request on every transient error without deduplication.

Question 458

The large online retailer is responding to a data protection inspection. A responsible-AI reviewer discovers this issue: A payment agent retries after a network disconnect and creates a duplicate purchase. What corrective step should the team take to restore the intended functionality most directly?

  1. Cancel pending deletion and present a reviewable action request
  2. Issue a fresh purchase request on every transient error without deduplication
  3. Simulate a dropped response and confirm exactly one committed purchase
  4. Reconcile existing transaction status before repeating execution

Explanation

Correct Answer

D:

The appropriate remediation is to reconcile existing transaction status before repeating execution, because the failure concerns idempotent agent actions: a payment agent retries after a network disconnect and creates a duplicate purchase. A meaningful verification should simulate a dropped response and confirm exactly one committed purchase, and the durable operating rule should require durable operation IDs and duplicate-transaction checks after the immediate fix.

Incorrect Answers

A:

This approach relates to human approval for destructive action: an agent could delete a business folder when asked to tidy files. If the team chose it instead, the targeted check to simulate a dropped response and confirm exactly one committed purchase could still fail because the underlying required action is to reconcile existing transaction status before repeating execution.

B:

During large online retailer remediation, the proposed step to issue a fresh purchase request on every transient error without deduplication is not the direct restoration action for idempotent agent actions when a payment agent retries after a network disconnect and creates a duplicate purchase; the keyed repair is reconcile existing transaction status before repeating execution. After remediation the team must simulate a dropped response and confirm exactly one committed purchase; the action in this option does not itself establish that target recovery outcome.

C:

The action to simulate a dropped response and confirm exactly one committed purchase can contribute to idempotent agent actions as a targeted acceptance check step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. During large online retailer remediation, the proposed step to simulate a dropped response and confirm exactly one committed purchase is not the direct restoration action for idempotent agent actions when a payment agent retries after a network disconnect and creates a duplicate purchase; the keyed repair is reconcile existing transaction status before repeating execution.

Question 459

A solution architect at a public library network wants the following issue to be prevented in later deployments: A payment agent retries after a network disconnect and creates a duplicate purchase. Which enduring control or engineering guardrail BEST prevents recurrence?

  1. Use idempotency keys for retryable side-effecting tool operations
  2. Reconcile existing transaction status before repeating execution
  3. Simulate a dropped response and confirm exactly one committed purchase
  4. Require durable operation IDs and duplicate-transaction checks
  5. Issue a fresh purchase request on every transient error without deduplication
  6. Validate audience expiry and least privilege on tokens

Explanation

Correct Answer

D:

The sustained prevention measure is to require durable operation IDs and duplicate-transaction checks. The implementation can still use the separate action to use idempotency keys for retryable side-effecting tool operations, but a guardrail constrains future changes. The public library network can measure whether the rule works by arranging to simulate a dropped response and confirm exactly one committed purchase.

Incorrect Answers

A:

The action to use idempotency keys for retryable side-effecting tool operations can contribute to idempotent agent actions as a implementation step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. For prevention at the public library network, the measure to use idempotency keys for retryable side-effecting tool operations is weaker or aimed at another issue than the idempotent agent actions guardrail to require durable operation IDs and duplicate-transaction checks needed when a payment agent retries after a network disconnect and creates a duplicate purchase.

B:

The action to reconcile existing transaction status before repeating execution can contribute to idempotent agent actions as a corrective remediation step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. For prevention at the public library network, the measure to reconcile existing transaction status before repeating execution is weaker or aimed at another issue than the idempotent agent actions guardrail to require durable operation IDs and duplicate-transaction checks needed when a payment agent retries after a network disconnect and creates a duplicate purchase.

C:

The action to simulate a dropped response and confirm exactly one committed purchase can contribute to idempotent agent actions as a targeted acceptance check step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. For prevention at the public library network, the measure to simulate a dropped response and confirm exactly one committed purchase is weaker or aimed at another issue than the idempotent agent actions guardrail to require durable operation IDs and duplicate-transaction checks needed when a payment agent retries after a network disconnect and creates a duplicate purchase.

E:

For prevention at the public library network, the measure to issue a fresh purchase request on every transient error without deduplication is weaker or aimed at another issue than the idempotent agent actions guardrail to require durable operation IDs and duplicate-transaction checks needed when a payment agent retries after a network disconnect and creates a duplicate purchase. The lasting guardrail for this specific threat is to require durable operation IDs and duplicate-transaction checks, not simply to complete an adjacent service task.

F:

Choosing to validate audience expiry and least privilege on tokens would direct effort toward external api auth for agent, whose characteristic issue is that an agent must access a customer API using delegated user permissions. The relevant corrective or preventive action here remains to require durable operation IDs and duplicate-transaction checks, with its own validation criteria.

Question 460

During a production incident, a machine-learning developer must accept a change for idempotent agent actions. The issue to be addressed is: A payment agent retries after a network disconnect and creates a duplicate purchase. Which concrete verification action demonstrates that the relevant behavior works?

  1. Use idempotency keys for retryable side-effecting tool operations
  2. Issue a fresh purchase request on every transient error without deduplication
  3. Ask for an unrelated privileged operation and confirm refusal or escalation
  4. Simulate a dropped response and confirm exactly one committed purchase

Explanation

Correct Answer

D:

To verify this change, the software consultancy should simulate a dropped response and confirm exactly one committed purchase. Operators still need the implementation to use idempotency keys for retryable side-effecting tool operations and the preventive requirement to require durable operation IDs and duplicate-transaction checks.

Incorrect Answers

A:

The action to use idempotency keys for retryable side-effecting tool operations can contribute to idempotent agent actions as a implementation step; here, a payment agent retries after a network disconnect and creates a duplicate purchase. In the software consultancy acceptance review, trying to use idempotency keys for retryable side-effecting tool operations does not test the exact idempotent agent actions condition where a payment agent retries after a network disconnect and creates a duplicate purchase; the relevant test is to simulate a dropped response and confirm exactly one committed purchase.

B:

In the software consultancy acceptance review, trying to issue a fresh purchase request on every transient error without deduplication does not test the exact idempotent agent actions condition where a payment agent retries after a network disconnect and creates a duplicate purchase; the relevant test is to simulate a dropped response and confirm exactly one committed purchase. Acceptance should instead demonstrate that the service can simulate a dropped response and confirm exactly one committed purchase, using the described failure as the test criterion.

C:

For agent goal definition, it can be sensible to ask for an unrelated privileged operation and confirm refusal or escalation if an agent attempts unrelated tasks because its service role is poorly scoped. Without the actual change to simulate a dropped response and confirm exactly one committed purchase, the answer lacks the correct operational scope even if the alternate action is technically valid elsewhere.

Developing AI Apps and Agents on Azure (AI-103) Practice Tests