Question 390 A dashboard treats a production monitoring safety score as proof that every response in the prior month is technically correct. Which verification is strongest for this particular failure mode? Inspect the monitored trace count and selection policy and disclose unscored traffic and judge limitations. Track scorer feedback by week and correlate rising unsupported claims with policy-version changes. Score factuality and harmful-language behavior separately on adversarial tone prompts and confirm safe alternatives pass. Confirm evaluation scores on the frozen set before production traffic exists, then validate live-trace scoring once deployed. ExplanationCorrect Answer A:Sampling gives evidence on observed traces, not exhaustive certification of unreviewed answers. Other Answers B:Inspect the monitored trace count and selection policy and disclose unscored traffic and judge limitations. C:Inspect the monitored trace count and selection policy and disclose unscored traffic and judge limitations. D:Inspect the monitored trace count and selection policy and disclose unscored traffic and judge limitations.
Question 391 A dashboard treats a production monitoring safety score as proof that every response in the prior month is technically correct. What change to engineering governance would reduce this risk over time? Preserve precise evaluation/monitoring provenance and prohibit claims of universal correctness from sampled metrics. Define separate pre-release quality gates and post-release monitoring thresholds. Schedule drift and freshness reviews from sampled live traces with approved retention boundaries. Maintain independent safety evaluation thresholds rather than collapsing every outcome into one overall correctness score. ExplanationCorrect Answer A:Sampling gives evidence on observed traces, not exhaustive certification of unreviewed answers. Other Answers B:That sampling gives evidence on observed traces, not exhaustive certification of unreviewed answers. C:Sampling gives evidence on observed traces, not exhaustive certification of unreviewed answers. D:This control addresses a different exposure: A customer-service assistant gives a helpful policy summary but uses harassing language toward the requester in its final paragraph. It would not prevent this failure, because sampling gives evidence on observed traces, not exhaustive certification of unreviewed answers.
Question 392 A business asks “What were last quarter's sales by region?” but a document-search agent retrieves narrative press releases instead of governed warehouse measures. Which engineering action most directly addresses this specific requirement? Route policy prose to the document agent and current counts to governed Genie data tooling, then reconcile the combined answer with citations. Log tool result identifiers and prompt context spans in the trace, then use recorded traces to review the failing branches. Route the structured analytics intent to a permissioned Genie data interface using the approved semantic definitions. Apply caller identity and row/document entitlements before every delegated tool call, including inter-agent handoffs. ExplanationCorrect Answer C:The action targets the stated cause: Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. Check the repair by: compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query. Other Answers A:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. B:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. D:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries.
Question 393 A business asks “What were last quarter's sales by region?” but a document-search agent retrieves narrative press releases instead of governed warehouse measures. What architectural constraint is most directly responsible? Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. Trace context provides causality between retrieval, tool execution and final generation; isolated logs cannot reconstruct the path. Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path. ExplanationCorrect Answer A:Route the structured analytics intent to a permissioned Genie data interface using the approved semantic definitions. Other Answers B:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. C:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. D:That mechanism describes another problem: A low-privilege employee asks a multi-agent supervisor to retrieve confidential payroll and the document agent forwards unrestricted snippets to the data agent. It does not explain this case, where business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries.
Question 394 A business asks “What were last quarter's sales by region?” but a document-search agent retrieves narrative press releases instead of governed warehouse measures. What validation evidence would most specifically confirm the resolution? Compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query. Select a failed response in MLflow and navigate to the exact source/tool span responsible for its unsupported content. Replay with an unauthorized identity and verify neither delegated tool trace contains payroll records. Test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance. ExplanationCorrect Answer A:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. Other Answers B:Compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query. C:Compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query. D:Compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query.
Question 395 A business asks “What were last quarter's sales by region?” but a document-search agent retrieves narrative press releases instead of governed warehouse measures. What prevention measure is most targeted to this underlying cause? Document per-agent evidence boundaries and validate cross-agent answer aggregation. Keep intent-routing tests for structured measures and enforce semantic-model governance. Require reproducible trace IDs and stable experiment tagging for each agent revision. Propagate least-privilege identity and audit access decisions across all agent boundaries. ExplanationCorrect Answer B:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. Other Answers A:Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. C:This control addresses a different exposure: A multi-tool agent is returning unpredictable answers because intermediate tool results are not linked to the generated text in its MLflow experiment. It would not prevent this failure, because business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. D:That business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries.
Question 396 A user asks a mixed question about warranty terms and current claim counts, but one general assistant fabricates both from a single paragraph. Which implementation decision best restores the intended behavior? Reconcile the March exam's Agent Framework terms with the October 2026 Agent Bricks CLI deployment guidance before migrating code. Apply caller identity and row/document entitlements before every delegated tool call, including inter-agent handoffs. Route the structured analytics intent to a permissioned Genie data interface using the approved semantic definitions. Route policy prose to the document agent and current counts to governed Genie data tooling, then reconcile the combined answer with citations. ExplanationCorrect Answer D:The action targets the stated cause: Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. Check the repair by: test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance. Other Answers A:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. B:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. C:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types.
Question 397 A user asks a mixed question about warranty terms and current claim counts, but one general assistant fabricates both from a single paragraph. Which underlying technical principle explains why this occurs? Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path. Historical certification vocabulary and current agent serving products can describe different deployment lifecycles. Business aggregate questions require trusted table/query semantics, not document similarity that may return stale marketing summaries. ExplanationCorrect Answer A:Route policy prose to the document agent and current counts to governed Genie data tooling, then reconcile the combined answer with citations. Other Answers B:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. C:That mechanism describes another problem: A custom Python agent uses an old deployment scaffold that no longer aligns with current Agent Runtime guidance. It does not explain this case, where different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. D:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types.
Question 398 A user asks a mixed question about warranty terms and current claim counts, but one general assistant fabricates both from a single paragraph. After correction, what targeted test should demonstrate success? Compare the answer to a validated SQL aggregate and inspect permissions and the executed Genie query. Verify a staging deployment with supported agent entry points, health checks and trace capture against current docs. Replay with an unauthorized identity and verify neither delegated tool trace contains payroll records. Test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance. ExplanationCorrect Answer D:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. Other Answers A:Test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance. B:Test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance. C:Test a mixed question and verify the policy citation and SQL-backed count both appear with correct provenance.
Question 399 A user asks a mixed question about warranty terms and current claim counts, but one general assistant fabricates both from a single paragraph. Which release safeguard most directly addresses this failure class? Propagate least-privilege identity and audit access decisions across all agent boundaries. Maintain a platform-version crosswalk and migration test suite rather than hard-coding legacy sample APIs. Document per-agent evidence boundaries and validate cross-agent answer aggregation. Keep intent-routing tests for structured measures and enforce semantic-model governance. ExplanationCorrect Answer C:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. Other Answers A:This control addresses a different exposure: A low-privilege employee asks a multi-agent supervisor to retrieve confidential payroll and the document agent forwards unrestricted snippets to the data agent. It would not prevent this failure, because different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. B:That different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types. D:Different facts live in different authoritative systems; a multi-agent router can delegate to each without conflating evidence types.
Question 400 A low-privilege employee asks a multi-agent supervisor to retrieve confidential payroll and the document agent forwards unrestricted snippets to the data agent. Which design change directly satisfies the stated constraints? Route the structured analytics intent to a permissioned Genie data interface using the approved semantic definitions. Apply caller identity and row/document entitlements before every delegated tool call, including inter-agent handoffs. Classify the incident as a privacy/access-control failure, block unauthorized retrieval and validate output against disclosure rules. Route policy prose to the document agent and current counts to governed Genie data tooling, then reconcile the combined answer with citations. ExplanationCorrect Answer B:The action targets the stated cause: Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path. Check the repair by: replay with an unauthorized identity and verify neither delegated tool trace contains payroll records. Other Answers A:Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path. C:Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path. D:Delegating to a specialist does not grant new privileges; cross-agent context is another potential disclosure path.
