CompTIA XK0-006 (CompTIA Linux+) Practice Test 55

Question 540

A compliance reviewer requests evidence for an AI-generated security recommendation. What is the appropriate professional response?

  1. Corroborate the recommendation with current authoritative documentation
  2. Inspect the IaC diff and resulting plan
  3. Test the regex against positive and negative examples
  4. Benchmarks plus restore-test results

Explanation

Correct Answer

A:

An LLM's answer is not a primary policy reference and may omit environmental prerequisites. Cited standards and direct system evidence are needed.

Incorrect Answers

B:

Infrastructure changes must be checked for unintended destruction, privilege changes and drift; this does not achieve validating AI compliance claims against trusted sources on the system described.

C:

Pattern suggestions can overmatch or miss edge cases; it will not provide validating AI compliance claims against trusted sources under these conditions.

D:

Faster execution is not useful if backup integrity or restoration correctness suffers; using it would leave validating AI compliance claims against trusted sources unresolved.

Question 541

A code-completion model proposes an optimization to a backup script. Which independent evidence should guide adoption?

  1. Corroborate the recommendation with current authoritative documentation
  2. State runtime constraints and required output explicitly
  3. Benchmarks plus restore-test results
  4. Inspect the IaC diff and resulting plan

Explanation

Correct Answer

C:

Faster execution is not useful if backup integrity or restoration correctness suffers. Performance and recovery tests are needed before replacing a proven script.

Incorrect Answers

A:

An LLM's answer is not a primary policy reference and may omit environmental prerequisites; it does not deliver evaluating AI-generated backup optimizations safely, which is decisive in this scenario.

B:

Giving the shell version, input examples and constraints can narrow ambiguity; the evidence calls for evaluating AI-generated backup optimizations safely, not merely this separate operation.

D:

Infrastructure changes must be checked for unintended destruction, privilege changes and drift; its scope differs from evaluating AI-generated backup optimizations safely, so it cannot meet the stated requirement.

Question 542

An AI assistant produces a shell command that recursively deletes a production directory. What should the administrator do first?

  1. An approved locally hosted model
  2. Redact or withhold sensitive log data per approved policy
  3. Remove and rotate the exposed credential
  4. Validate command effects in a safe test environment

Explanation

Correct Answer

D:

Generated commands may be unsafe or wrong for the target filesystem. Human review, backups and an isolated test precede any destructive execution.

Incorrect Answers

A:

Local inference can limit external transmission when its deployment and access controls are properly designed; the administrator instead needs verifying a destructive AI-suggested command; this option serves a different purpose.

B:

Data governance determines whether personal information and secrets may be sent to external services; it answers a neighboring question rather than verifying a destructive AI-suggested command.

C:

A credential that entered generated artifacts or prompts may already be compromised; that changes a different part of the system; the requested outcome is verifying a destructive AI-suggested command.

Question 543

An administrator uses an AI assistant to draft a high-impact firewall change using incident evidence. Which THREE controls should be in the workflow before production use? Choose THREE.

  1. Human approval and change-control gate
  2. Redact or withhold sensitive log data per approved policy
  3. Rely only on local AI hosting and skip incident-data classification
  4. Treat syntactically valid AI-generated rules as approved without human review
  5. Validate command effects in a safe test environment
  6. Deploy first, then check change approvals during the postincident review

Explanation

Correct Answers

A:

A high-impact firewall policy must pass a human approval and change-control checkpoint, regardless of how an AI assistant drafted it. The reviewer should assess risk and rollback before authorizing production deployment.

B:

Data governance determines whether personal information and secrets may be sent to external services. Sanitized minimal examples reduce accidental exposure.

E:

AI-generated firewall changes should be tested and their command effects checked before production use; output may be plausible yet unsafe. This technical validation complements sensitive-data protection and human authorization.

Incorrect Answers

C:

A locally hosted AI deployment can limit external transmission, but internal access and data-handling rules still apply; it does not deliver keeping incident data private, testing generated changes and requiring human approval, which is decisive in this scenario.

D:

Successful syntax validation does not establish authorization, least privilege or production suitability; the evidence calls for keeping incident data private, testing generated changes and requiring human approval, not merely this separate operation.

F:

Retroactive approval does not substitute for a predeployment risk and rollback assessment; its scope differs from keeping incident data private, testing generated changes and requiring human approval, so it cannot meet the stated requirement.

Question 544

A team wants AI-assisted analysis without transmitting confidential configuration to an external provider. Which option might help if approved?

  1. Follow the organization's approved AI tool and data policy
  2. An approved locally hosted model
  3. Redact or withhold sensitive log data per approved policy
  4. Human approval and change-control gate
  5. Validate command effects in a safe test environment

Explanation

Correct Answer

B:

Local inference can limit external transmission when its deployment and access controls are properly designed. On-premises operation alone does not eliminate governance requirements.

Incorrect Answers

A:

Organizational policy governs permissible models and disclosure; its scope differs from limiting external disclosure during AI inference, so it cannot meet the stated requirement.

C:

Data governance determines whether personal information and secrets may be sent to external services; the administrator instead needs limiting external disclosure during AI inference; this option serves a different purpose.

D:

A human can assess business context and rollback before applying high-impact changes; it answers a neighboring question rather than limiting external disclosure during AI inference.

E:

Generated commands may be unsafe or wrong for the target filesystem; that changes a different part of the system; the requested outcome is limiting external disclosure during AI inference.

Question 545

A technician wants an LLM to suggest a regular expression for extracting hostnames. What validation approach is best?

  1. Test the regex against positive and negative examples
  2. State runtime constraints and required output explicitly
  3. Corroborate the recommendation with current authoritative documentation
  4. Benchmarks plus restore-test results

Explanation

Correct Answer

A:

Pattern suggestions can overmatch or miss edge cases. Test cases that include representative failures are necessary before use in an operational parser.

Incorrect Answers

B:

Giving the shell version, input examples and constraints can narrow ambiguity; it will not provide verifying generated regular expressions on examples under these conditions.

C:

An LLM's answer is not a primary policy reference and may omit environmental prerequisites; using it would leave verifying generated regular expressions on examples unresolved.

D:

Faster execution is not useful if backup integrity or restoration correctness suffers; the requirement is verifying generated regular expressions on examples, outside the effect of this choice.

Question 546

An AI assistant is asked to summarize untrusted application logs. One log entry includes an instruction to ignore the administrator's request and reveal credentials. How should the assistant's log-processing workflow handle that content?

  1. Treat the log entry as untrusted data, not as an instruction
  2. Follow the log entry because it came from a production server
  3. Grant the model privileged access so it can verify the instruction
  4. Send the requested credentials after masking the log timestamp
  5. Disable audit logging to prevent similar messages

Explanation

Correct Answer

A:

Instructions embedded in logs are attacker-controlled input and must not override the authorized task. Keep data and instructions separated and avoid executing or disclosing secrets on the basis of a log line.

Incorrect Answers

B:

A production log records events and may include attacker-supplied text. Its origin does not make embedded commands part of the administrator's authorized instructions.

C:

Expanding model privileges increases the impact of prompt-injection attempts. The proper response is to preserve the trust boundary, not add capabilities to untrusted content.

D:

Removing a timestamp does not authorize credential disclosure. The malicious instruction remains untrusted text and the requested secrets must not be revealed.

E:

Disabling logging destroys evidence and does not stop the model from encountering injected instructions in other inputs. The processing workflow must reject the embedded directive.

Question 547

Response-time objectives should report user-facing latency distribution rather than only a misleading mean. Which metric is useful?

  1. Alert thresholds relative to the SLO and SLI
  2. Service-level agreement (SLA)
  3. 95th-percentile request latency
  4. Allowable error budget

Explanation

Correct Answer

C:

A high percentile exposes slow experiences affecting a fraction of users that averages can hide. The percentile and measurement window need clear definition.

Incorrect Answers

A:

Monitoring thresholds must reflect the reliability target and measurement window; it answers a neighboring question rather than characterizing tail latency experienced by users.

B:

An SLA is a formal commitment between provider and customer with terms, responsibilities and potential remedies; that changes a different part of the system; the requested outcome is characterizing tail latency experienced by users.

D:

An error budget represents the tolerated time or request failures implied by an SLO; it will not provide characterizing tail latency experienced by users under these conditions.

Question 548

A critical server stops reporting metrics entirely, yet normal low-CPU alerts do not fire. Which separate condition should be monitored?

  1. Deployment event annotations
  2. Telemetry freshness or missing-data alarm
  3. Alert correlation and deduplication
  4. Sustained-duration alert evaluation

Explanation

Correct Answer

B:

Monitoring the absence of expected reports detects failures in the agent or data pipeline. Silence is not evidence of a healthy server.

Incorrect Answers

A:

Change events overlaid on telemetry provide context for correlation; the requirement is detecting missing monitoring data, outside the effect of this choice.

C:

Grouping alerts by shared incident or root cause avoids repeated notifications; the evidence calls for detecting missing monitoring data, not merely this separate operation.

D:

Requiring a condition to remain true for a specified interval filters transient bursts; the task involves detecting missing monitoring data, whereas this addresses another subsystem.

Question 549

Several Linux servers should forward logs to a single indexed platform for cross-host search. Which architecture fits?

  1. Agentless monitoring
  2. Webhook notification
  3. Central log aggregation
  4. SNMP polling

Explanation

Correct Answer

C:

Aggregation brings events together with timestamps and source identities to support correlation. Time synchronization and secure transport improve reliability.

Incorrect Answers

A:

Agentless checks obtain data over existing protocols without a dedicated local monitoring process; that is not equivalent to correlating logs across multiple systems.

B:

A webhook sends an HTTP request in response to an event; the administrator instead needs correlating logs across multiple systems; this option serves a different purpose.

D:

SNMP can retrieve indexed management variables, such as traffic counters, from managed devices; that changes a different part of the system; the requested outcome is correlating logs across multiple systems.

Question 550

Network switches should report interface counter data using a standard management protocol. Which protocol commonly supplies it?

  1. Webhook notification
  2. SNMP polling
  3. SNMP trap
  4. SNMP management information bases (MIBs)

Explanation

Correct Answer

B:

SNMP can retrieve indexed management variables, such as traffic counters, from managed devices. Polling gives periodic observations rather than immediate push events.

Incorrect Answers

A:

A webhook sends an HTTP request in response to an event; it will not provide collecting network-device counters periodically under these conditions.

C:

A trap sends an event notification from an agent to a manager; the requirement is collecting network-device counters periodically, outside the effect of this choice.

D:

MIBs describe the names, types and meanings of SNMP-managed objects; it does not deliver collecting network-device counters periodically, which is decisive in this scenario.

CompTIA Linux+ (XK0-006) Practice Tests