Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 60

Question 590

An operational risk could arise because a shared MCP server might honor an unauthorized tool invocation. The AI system supports the business need that a Copilot Studio agent needs a reusable interface to governed operations. Which guardrail is designed for this threat?

  1. Restrict available actions and require confirmation for consequential changes
  2. Preserve table and row access controls for retrieval
  3. Narrow allowed tool capabilities and verify server-side authorization
  4. Enforce business-unit and row-level permissions at the data layer

Explanation

Correct Answer

C:

In the context where a Copilot Studio agent needs a reusable interface to governed operations, the exposure is that a shared MCP server might honor an unauthorized tool invocation; containing it requires a guardrail in the operation where MCP presents supported tools and resources through a common integration contract.

Incorrect Answers

A:

This restriction is associated with employees express several intents and need interactive clarification and addresses the exposure that a conversational tool could invoke consequential changes from an ambiguous request; it does not specifically protect the target workflow against a shared MCP server might honor an unauthorized tool invocation.

B:

The proposed control targets Dataverse retrieval might reveal a row not authorized to the caller, while the given risk scenario is a shared MCP server might honor an unauthorized tool invocation; these exposures arise in different operational contexts.

D:

The alternate boundary mitigates shared data tables might expose rows to a different business unit, which is relevant to agents and apps need shared definitions of customers and work orders; it leaves the current workflow vulnerable to a shared MCP server might honor an unauthorized tool invocation.

Question 591

Troubleshooting starts from a business complaint: Global averages hide a regional escalation spike. The affected process is: A conversational change affects only one support geography. Which investigation is most directly tied to the reported defect?

  1. Compare outcomes by locale, channel and agent version
  2. Compare target environment version and active channel metadata
  3. Diff tool traces and evaluation outcomes against baseline
  4. Verify spans and errors arrive in the expected telemetry resource

Explanation

Correct Answer

A:

For the case where a conversational change affects only one support geography, the diagnostic path must explain the failure that global averages hide a regional escalation spike; comparing outcomes by locale, channel and agent version targets the mechanism by which cohort telemetry supports safe staged validation.

Incorrect Answers

B:

For the failure that deployment checks stop at package import success, comparing target environment version and active channel metadata is relevant; this question instead reports that global averages hide a regional escalation spike, so different evidence is needed.

C:

New planning behavior occasionally leaks internal notes; examining that other path does not determine whether cohort telemetry supports safe staged validation, which is what distinguishes the issue in the stated use case.

D:

Verifying spans and errors arrive in the expected telemetry resource explores a separate connection, where connected observability supplies traces and operational metrics; the diagnosed requirement is a conversational change affects only one support geography, and the actual defect is global averages hide a regional escalation spike.

Question 592

A staging test has to detect the specific defect that a failing agent shows only a generic HTTP error. In the workflow where operations need to correlate prompts, model calls and tools in a Foundry workflow, which exercise verifies the corrective behavior?

  1. Use an annotated query with a known authoritative paragraph
  2. Reproduce the intent using a captured test trace
  3. Inject a tool timeout and locate the failing span
  4. Inject delay into one dependency and locate the timing change

Explanation

Correct Answer

C:

Injecting a tool timeout and locating the failing span challenges the path where agent trace spans provide component-level execution evidence through connected telemetry for the business setting operations need to correlate prompts, model calls and tools in a Foundry workflow; the specific defect to exclude is that a failing agent shows only a generic HTTP error.

Incorrect Answers

A:

In a solution where an assistant responds fluently but cites unrelated support articles, using an annotated query with a known authoritative paragraph could catch the defect that surface quality scores appear healthy despite wrong documents; the current release gate must challenge the failure where a failing agent shows only a generic HTTP error.

B:

In the alternate setting where the planner calls an order-cancel tool for status requests, the test exercises step traces expose planning, selected tools and arguments; even a pass would not show that a failing agent shows only a generic HTTP error has been prevented.

D:

Injecting delay into one dependency and locating the timing change verifies an adjacent objective where agents regularly exceed an interactive response-time objective; it leaves untested the causal path in which agent trace spans provide component-level execution evidence through connected telemetry for this case.

Question 593

A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: Customers ask for current delivery or payment status. Known defect: A cached knowledge answer is several hours old. Which KPI should be selected?

  1. Live status answers matching system-of-record state
  2. Unsupported answer rate during retrieval failures
  3. Authorized tool calls completed without excess privileges
  4. Share of material claims supported by retrievable citations

Explanation

Correct Answer

A:

Because authorized actions query the system of record at request time, the outcome worth following is live status answers matching system-of-record state; the metric is tied to the reported issue—a cached knowledge answer is several hours old—for the scenario where customers ask for current delivery or payment status.

Incorrect Answers

B:

For the situation where required knowledge stores are intermittently unreachable, unsupported answer rate during retrieval failures is a sensible indicator; the problem that a cached knowledge answer is several hours old requires measurement of live status answers matching system-of-record state.

C:

That scorecard concerns the activity in which a Copilot Studio agent needs a reusable interface to governed operations and its result of authorized tool calls completed without excess privileges; the reported shortcoming here is a cached knowledge answer is several hours old.

D:

Although share of material claims supported by retrievable citations is observable, the affected process involves customers ask for current delivery or payment status; its acceptance outcome is live status answers matching system-of-record state, not the alternative measurement.

Question 594

An incident review has one confirmed symptom: Regression appears only for compound user requests. For the use case where a new topic description changes which agent is invoked, which targeted diagnostic check would reveal the most pertinent evidence?

  1. Compare cost per accepted output across model versions
  2. Analyze failures by paraphrase group and boundary type
  3. Audit token use and secret inventory for agent services
  4. Diff representative before-and-after activity maps
  5. Inspect model and tool spans under one operation identifier

Explanation

Correct Answer

D:

For the situation where a new topic description changes which agent is invoked, the defect is that regression appears only for compound user requests; diffing representative before-and-after activity maps helps verify the cause-and-effect path: trace comparisons reveal shifts in plan selection and step order.

Incorrect Answers

A:

Comparing cost per accepted output across model versions explores a separate connection, where joint quality and consumption measures capture real efficiency; the diagnosed requirement is a new topic description changes which agent is invoked, and the actual defect is regression appears only for compound user requests.

B:

Slightly different wording breaks correct entity selection; examining that other path does not determine whether trace comparisons reveal shifts in plan selection and step order, which is what distinguishes the issue in the stated use case.

C:

The evidence this method produces applies to a workflow where a backend agent must call an approved Azure service without embedded secrets; it cannot by itself establish why regression appears only for compound user requests in the current scenario.

E:

Inspecting model and tool spans under one operation identifier explores a separate connection, where agent trace spans provide component-level execution evidence through connected telemetry; the diagnosed requirement is a new topic description changes which agent is invoked, and the actual defect is regression appears only for compound user requests.

Question 595

A policy review covers an agent for which solution import succeeds but users still see the previous behavior. Its most relevant exposure is that an imported but unpublished agent might cause users to see stale behavior. What safeguard should be required before release?

  1. Revoke shared credentials and forbid cross-stage access
  2. Roll back or withhold publication if acceptance checks fail
  3. Stage compatible transitions and protect authoritative fields
  4. Disable the candidate for a cohort on critical regression

Explanation

Correct Answer

B:

Solution import succeeds but users still see the previous behavior; its risk scenario is that an imported but unpublished agent might cause users to see stale behavior; the safeguard fits because post-import validation includes publishing and channel behavior and the restriction addresses that specific exposure.

Incorrect Answers

A:

Another process could need this guardrail because staging credentials might allow writes to production business systems; the design being assessed supports solution import succeeds but users still see the previous behavior and has the explicit exposure that an imported but unpublished agent might cause users to see stale behavior.

C:

A customer handoff depends on shared record and term mappings; its guardrail addresses the possibility that cross-Dynamics releases might overwrite authoritative customer fields; the present architecture must constrain the different threat that an imported but unpublished agent might cause users to see stale behavior.

D:

A conversational change affects only one support geography; its guardrail addresses the possibility that a regional pilot might produce a severe local regression hidden by global averages; the present architecture must constrain the different threat that an imported but unpublished agent might cause users to see stale behavior.

Question 596

An incident review has one confirmed symptom: A custom service would duplicate built-in host orchestration. For the use case where employees want curated assistance without owning a separate LLM runtime, which targeted diagnostic check would reveal the most pertinent evidence?

  1. Compare prompt versions used by dependent apps
  2. Review action inputs, examples and schema-parse logs
  3. Review agent capability manifest against requested functions
  4. Compare advertised tools with backend permissions

Explanation

Correct Answer

C:

For the situation where employees want curated assistance without owning a separate LLM runtime, the defect is that a custom service would duplicate built-in host orchestration; reviewing agent capability manifest against requested functions helps verify the cause-and-effect path: declarative agents define instructions, knowledge and actions in the supported host.

Incorrect Answers

A:

The proposed inspection is associated with multiple low-code apps need a common extraction prompt, where centralized reusable AI capabilities promote consistency and controlled updates; it would not expose the failure mode that a custom service would duplicate built-in host orchestration with the required specificity.

B:

The proposed inspection is associated with an agent must turn service notes into a fixed JSON summary, where prompt actions encapsulate reusable inputs and expected outputs; it would not expose the failure mode that a custom service would duplicate built-in host orchestration with the required specificity.

D:

The proposed probe concerns several agents need the same inventory query operation, where a supported MCP server exposes typed tool capabilities to clients; this incident instead requires evidence about declarative agents define instructions, knowledge and actions in the supported host and the failure that a custom service would duplicate built-in host orchestration.

Question 597

A solution architecture is being reviewed for this business situation: External credit service responds slowly during regional outages. The observed defect is that a sales workflow hangs awaiting optional enrichment. Which architectural decision addresses the underlying requirement?

  1. Design timeout and degraded-mode handling for Sales enrichments
  2. Use a governed connector for live external account data
  3. Index governed external content using Microsoft 365 Copilot connectors
  4. Add approved case-summary guidance to the service workflow

Explanation

Correct Answer

A:

The defect is that a sales workflow hangs awaiting optional enrichment; the expected business case is that external credit service responds slowly during regional outages; the design fits because bounded timeouts and graceful fallback preserve essential user work.

Incorrect Answers

B:

When sellers need current credit exposure held outside CRM, a supported connector can retrieve external information with explicit identity mapping; in the stated workflow, however, a sales workflow hangs awaiting optional enrichment; the relevant architectural behavior is that bounded timeouts and graceful fallback preserve essential user work.

C:

The technical behavior of this alternative is that supported connectors expose external content with discoverability and security metadata; the user case being assessed is external credit service responds slowly during regional outages with a different failure: a sales workflow hangs awaiting optional enrichment.

D:

The technical behavior of this alternative is that tailored instructions and contextual fields standardize case handoff content; the user case being assessed is external credit service responds slowly during regional outages with a different failure: a sales workflow hangs awaiting optional enrichment.

Question 598

The fault affects an AI-powered workflow in which a support agent changes how it classifies priority intents. Specifically, high-severity tickets begin routing to general queues. Which diagnostic investigation best distinguishes the relevant failure path?

  1. Compare deployed agent configuration to its approved manifest
  2. Diff tool traces and evaluation outcomes against baseline
  3. Compare ticket assignment confusion matrices across versions
  4. Inspect imported connection references and environment variables

Explanation

Correct Answer

C:

For the case where a support agent changes how it classifies priority intents, the diagnostic path must explain the failure that high-severity tickets begin routing to general queues; comparing ticket assignment confusion matrices across versions targets the mechanism by which controlled service evaluations validate routing before rollout.

Incorrect Answers

A:

In the process where model and tool changes are promoted independently, the diagnostic procedure could explain a release uses stale prompts against a new tool signature; this particular incident instead turns on whether controlled service evaluations validate routing before rollout and why high-severity tickets begin routing to general queues.

B:

In the process where a hosted agent gains an additional research tool, the diagnostic procedure could explain new planning behavior occasionally leaks internal notes; this particular incident instead turns on whether controlled service evaluations validate routing before rollout and why high-severity tickets begin routing to general queues.

D:

For the failure that agent writes real customer data into the sandbox, inspecting imported connection references and environment variables is relevant; this question instead reports that high-severity tickets begin routing to general queues, so different evidence is needed.

Question 599

Support engineers report this failure: Each agent builds a different unaudited HTTP wrapper. The business context is: Several agents need the same inventory query operation. What evidence should be collected to isolate the cause?

  1. Compare advertised tools with backend permissions
  2. Inspect duplicate request IDs across retries
  3. Compare output format across supported channels
  4. Inspect transfer payloads for missing or overexposed data

Explanation

Correct Answer

A:

Each agent builds a different unaudited HTTP wrapper; to attribute the defect rather than guess at a fix, comparing advertised tools with backend permissions examines the dependency where a supported MCP server exposes typed tool capabilities to clients.

Incorrect Answers

B:

That probe evaluates idempotency keys tie repeated submissions to one business operation in a workflow where network retries can repeat a single purchase request; the reported failure concerns the path for which a supported MCP server exposes typed tool capabilities to clients instead.

C:

The proposed inspection is associated with the same agent is offered in voice and asynchronous chat, where channel-aware response design adapts wording and modality; it would not expose the failure mode that each agent builds a different unaudited HTTP wrapper with the required specificity.

D:

The evidence this method produces applies to a workflow where a voice bot cannot resolve a complex eligibility dispute; it cannot by itself establish why each agent builds a different unaudited HTTP wrapper in the current scenario.

Question 600

Support engineers report this failure: Lower-cost routing degrades a business-critical subgroup. The business context is: Average accuracy hides failures on complex finance questions. What evidence should be collected to isolate the cause?

  1. Compare quality by task complexity and selected model
  2. Correlate request rates with throttles and queue depth
  3. Audit training-data writers and model approvers
  4. Compare ticket assignment confusion matrices across versions

Explanation

Correct Answer

A:

In the setting where average accuracy hides failures on complex finance questions, comparing quality by task complexity and selected model can isolate the failure in which lower-cost routing degrades a business-critical subgroup, because stratified evaluation guides thresholds for difficult requests.

Incorrect Answers

B:

A regional launch multiplies concurrent agent requests; its matching diagnostic check is reasonable for rate limiting causes intermittent failures; the present investigation needs evidence for the different failure where lower-cost routing degrades a business-critical subgroup.

C:

The proposed probe concerns business analysts can submit examples but not change production behavior, where separated data-curation and release permissions preserve integrity; this incident instead requires evidence about stratified evaluation guides thresholds for difficult requests and the failure that lower-cost routing degrades a business-critical subgroup.

D:

That probe evaluates controlled service evaluations validate routing before rollout in a workflow where a support agent changes how it classifies priority intents; the reported failure concerns the path for which stratified evaluation guides thresholds for difficult requests instead.

Agentic AI Business Solutions Architect (AB-100) Practice Tests