McAfee Secure

Cisco 350-901 Bundle

Exam Code: 350-901

Exam Name Designing, Deploying, and Managing Network Automation Systems

Certification Provider: Cisco

Corresponding Certification: DevNet Professional

Cisco 350-901 Bundle $25.00

100% Cisco 350-901 Guaranteed Success With Testking Training Materials

Get Real 350-901 Exam Questions & Expert Verified Answers as Experienced on the Real Exam!

  • Questions & Answers

    350-901 Questions & Answers

    604 Questions & Answers

    The ultimate exam preparation tool, Questions & Answers mirrors the actual 350-901 exam while allowing you to hone and measure every aspect of your test-taking ability.

  • 350-901 Video Course

    350-901 Training Course

    106 Video Lectures

    350-901 Video Course is developed by Cisco Professionals to help you pass the 350-901 exam.

    Description

    This course will improve your knowledge and skills required to pass Designing, Deploying, and Managing Network Automation Systems exam.

Designing, Deploying, and Managing Network Automation Systems (350-901) Practice Tests

Satisfaction Guaranteed

Money Back Guarantee

Testking provides hassle-free money back guarantee with our products. That is because we have 100% trust in the abilities of our professional and experience product team, and our record is a proof of that.

99.6% PASS RATE
Total Cost: $164.98
Bundle Price: $139.98

Purchase Individually

  • Questions & Answers

    Questions & Answers

    604 Questions

    $124.99
  • 350-901 Video Course

    Training Course

    106 Video Lectures

    $39.99

Cisco 350-901 Practice Test 52

Question 510

At fictional Seabright Learning, an acceptance review of a reproducible simulation finds: A deployment stage targets all routers at once despite an approved plan to canary one site before fleet rollout. What technical evidence should be required before closing this change?

  1. A fresh isolated runner installs the dependency from the manifest and completes the build without relying on developer machine state.
  2. GitLab deployment records show the canary passed before other sites changed, and credentials cannot access unrelated workspaces.
  3. The working tree reports the intended revision and no uncommitted cross-branch edits before test execution.
  4. A recreated container reads the expected persistent audit or inventory records from the configured volume.

Explanation

Correct Answer

B:

GitLab deployment records show the canary passed before other sites changed, and credentials cannot access unrelated workspaces. This directly verifies the related correction: Deploy through bounded environment-scoped jobs with approved service credentials and staged canary or batch rollout. It checks the technical distinction: A deployment stage should enforce release scope and approval, not merely execute a command with unrestricted infrastructure access. Applied here, the relevant facts are: A deployment stage targets all routers at once despite an approved plan to canary one site before fleet rollout. The question asks what technical evidence should be required before closing this change.

Incorrect Answers

A:

That test belongs to another control with this basis: This directly verifies the related correction: Declare and lock the required dependency in the build manifest and install it in the pipeline's reproducible environment. In the current case, the governing reason is: This directly verifies the related correction: Deploy through bounded environment-scoped jobs with approved service credentials and staged canary or batch rollout. The direct verification is “GitLab deployment records show the canary passed before other sites changed, and credentials cannot access unrelated workspaces”.

C:

The alternative verifies a different condition: This directly verifies the related correction: Safely stash or commit unrelated work and check out the intended branch or detached revision in a clean worktree. The current case instead depends on: This directly verifies the related correction: Deploy through bounded environment-scoped jobs with approved service credentials and staged canary or batch rollout. Confirm that “GitLab deployment records show the canary passed before other sites changed, and credentials cannot access unrelated workspaces”.

D:

That test belongs to another control with this basis: This directly verifies the related correction: Mount an appropriate persistent named volume or approved bind mount for data that must survive container recreation. In the current case, the governing reason is: This directly verifies the related correction: Deploy through bounded environment-scoped jobs with approved service credentials and staged canary or batch rollout. The direct verification is “GitLab deployment records show the canary passed before other sites changed, and credentials cannot access unrelated workspaces”.

Question 511

At fictional Seabright Learning, an acceptance review of a reproducible simulation finds: A network automation test job installs the latest REST API client and breaks a method signature used by the script. Which explanation best matches the actual broken dependency?

  1. Git can retrieve individual tracked paths from a commit; selecting file scope avoids broad branch-level rollback.
  2. A cherry-picked commit does not automatically bring every ancestor or dependent change from its original branch.
  3. Unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged.
  4. A reliable source-of-truth workflow separates approved intent from observed device state and defines how changes flow between them.

Explanation

Correct Answer

C:

Unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged. The corresponding repair is: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. Acceptance should show: The runner resolves the reviewed versions and integration tests pass consistently across repeated clean builds. Applied here, the relevant facts are: A network automation test job installs the latest REST API client and breaks a method signature used by the script. The question asks which explanation best matches the actual broken dependency.

Incorrect Answers

A:

This alternative is a response to another technical cause: The corresponding repair is: Restore only the specifically reviewed historical file from the desired revision instead of replacing the whole branch state. The reported case instead depends on: The corresponding repair is: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. The action that directly addresses it is “Unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged”.

B:

The proposed action belongs to the mechanism: The corresponding repair is: Identify the prerequisite commits and cherry-pick a minimal compatible sequence or adapt the patch for the target baseline. Here the stated fault follows a different mechanism: The corresponding repair is: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. The corrective choice is “Unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged”.

D:

This alternative is a response to another technical cause: The corresponding repair is: Read authorized desired state from the designated source of truth and reconcile observed drift through an approved change path. The reported case instead depends on: The corresponding repair is: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. The action that directly addresses it is “Unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged”.

Question 512

During an AI recommendation accuracy study at fictional Amberlight Energy, the change record states: A developer sends internal Cisco API bearer tokens and customer inventory records to an unapproved hosted LLM to generate a Terraform module. Which control best prevents this particular failure from recurring?

  1. Add explicit timeouts, result freshness metadata and partial-failure contract tests to every MCP tool.
  2. Adopt AI data-classification rules, approved endpoints and automatic prompt redaction checks for developer workflows.
  3. Implement tool categorization, approval records and run-as identities for every privileged network agent operation.
  4. Require deterministic schema validation, diff preview and affirmative approval for any mutating conversational action.

Explanation

Correct Answer

B:

Adopt AI data-classification rules, approved endpoints and automatic prompt redaction checks for developer workflows. This prevents the recurrence because aI-assisted code tools may process submitted data outside the intended trust boundary, creating privacy and intellectual-property exposure. A matching immediate correction is: Sanitize inputs and use an approved private or contractually controlled AI service without exposing secrets or regulated network data.

Incorrect Answers

A:

The suggested review tracks a different technical risk: tool wrappers must preserve the meaning of missing, stale and failed data for the consumer agent. Here the sustained control must address: aI-assisted code tools may process submitted data outside the intended trust boundary, creating privacy and intellectual-property exposure. Choose “Adopt AI data-classification rules, approved endpoints and automatic prompt redaction checks for developer workflows”.

C:

This routine guards against another mechanism: agent tool chaining can enlarge the blast radius of an ambiguous user instruction unless execution permissions are independently gated. The recurring exposure in the current case is: aI-assisted code tools may process submitted data outside the intended trust boundary, creating privacy and intellectual-property exposure. The matching preventive check is “Adopt AI data-classification rules, approved endpoints and automatic prompt redaction checks for developer workflows”.

D:

The suggested review tracks a different technical risk: lLMs are useful for language interpretation but the execution boundary should accept validated structured intents, not arbitrary prose. Here the sustained control must address: aI-assisted code tools may process submitted data outside the intended trust boundary, creating privacy and intellectual-property exposure. Choose “Adopt AI data-classification rules, approved endpoints and automatic prompt redaction checks for developer workflows”.

Question 513

At fictional Seabright Learning, an acceptance review of a reproducible simulation finds: A network automation test job installs the latest REST API client and breaks a method signature used by the script. Which sustainable preventive control most directly avoids a recurrence?

  1. Require merge-request review and parse or lint validation after conflicts in network automation files.
  2. Use reproducible lockfiles, container digest pinning and scheduled compatibility tests before updating dependencies.
  3. Require clean-worktree status checks when CI or operators switch network automation release branches.
  4. Add explicit GitLab stage dependencies and enforce nonoptional predeploy validation checks.

Explanation

Correct Answer

B:

Use reproducible lockfiles, container digest pinning and scheduled compatibility tests before updating dependencies. This prevents the recurrence because unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged. A matching immediate correction is: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. Applied here, the relevant facts are: A network automation test job installs the latest REST API client and breaks a method signature used by the script. The question asks which sustainable preventive control most directly avoids a recurrence.

Incorrect Answers

A:

That safeguard is relevant to: git cannot automatically reconcile overlapping divergent edits; a human or reviewed tool must choose the resulting content before merge completion. The current failure instead follows from: unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged. Its prevention measure is “Use reproducible lockfiles, container digest pinning and scheduled compatibility tests before updating dependencies”.

C:

The suggested review tracks a different technical risk: git checkout can change branch and working-tree files; uncommitted modifications must be managed to avoid accidental mixing. Here the sustained control must address: unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged. Choose “Use reproducible lockfiles, container digest pinning and scheduled compatibility tests before updating dependencies”.

D:

That safeguard is relevant to: prevalidation is a safety boundary before changing network state, separate from checks after a change completes. The current failure instead follows from: unbounded dependency upgrades can introduce incompatible APIs and nondeterministic CI failures even when repository code is unchanged. Its prevention measure is “Use reproducible lockfiles, container digest pinning and scheduled compatibility tests before updating dependencies”.

Question 514

At fictional Cedarbridge Care Systems, during a campus API automation rollout, the operator observes: An Ansible network job uses a raw shell command to append the same interface description on every run, producing duplicate configuration lines after retries. Which observation would confirm that this exact failure is resolved?

  1. The device accepts the configured YANG node and a GET of the corresponding resource returns the intended model-shaped value.
  2. The job records successes and failures by device, retries only supported transient cases, and does not claim full success when any target fails.
  3. The integration refreshes an expired token once when permitted, does not leak credentials and reports genuine 403 denials accurately.
  4. A second playbook run reports no unnecessary changes, and captured device configuration contains one intended interface setting on each target.

Explanation

Correct Answer

D:

A second playbook run reports no unnecessary changes, and captured device configuration contains one intended interface setting on each target. This directly verifies the related correction: Replace non-idempotent shell appends with the platform-aware Ansible network configuration module and explicit desired interface state. It checks the technical distinction: Ansible's desired-state-aware network modules compare declared configuration with device state before determining required changes. Applied here, the relevant facts are: An Ansible network job uses a raw shell command to append the same interface description on every run, producing duplicate configuration lines after retries. The question asks which observation would confirm that this exact failure is resolved.

Incorrect Answers

A:

This observation would test a different technical dependency: This directly verifies the related correction: Build the request body from the correct YANG data node and use application/yang-data+json with valid RESTCONF resource paths. The current issue is governed by: This directly verifies the related correction: Replace non-idempotent shell appends with the platform-aware Ansible network configuration module and explicit desired interface state. Its decisive check is “A second playbook run reports no unnecessary changes, and captured device configuration contains one intended interface setting on each target”.

B:

The alternative verifies a different condition: This directly verifies the related correction: Handle per-target HTTP, parsing and retryable exceptions and persist a failure summary without misreporting completed targets. The current case instead depends on: This directly verifies the related correction: Replace non-idempotent shell appends with the platform-aware Ansible network configuration module and explicit desired interface state. Confirm that “A second playbook run reports no unnecessary changes, and captured device configuration contains one intended interface setting on each target”.

C:

That test belongs to another control with this basis: This directly verifies the related correction: Use the supported token refresh flow with secret storage and distinguish unauthenticated 401 from authorized-but-forbidden 403 responses. In the current case, the governing reason is: This directly verifies the related correction: Replace non-idempotent shell appends with the platform-aware Ansible network configuration module and explicit desired interface state. The direct verification is “A second playbook run reports no unnecessary changes, and captured device configuration contains one intended interface setting on each target”.

Question 515

While investigating a reproducible simulation for fictional Windward Maritime, the team reports: An Ansible network workflow deploys source directly from a mutable branch without a versioned build artifact. What acceptance check would prove the intended behavior was restored?

  1. The postvalidation stage confirms routes, adjacency and connectivity or triggers the documented recovery workflow with evidence.
  2. The merged file has no conflict markers, passes schema tests and retains the intended requirements from both branches.
  3. The pipeline consistently reaches a ready test topology and records the versioned lab definition and run results.
  4. The deployed artifact checksum matches the tested build artifact and the pipeline records its originating commit.

Explanation

Correct Answer

D:

The deployed artifact checksum matches the tested build artifact and the pipeline records its originating commit. This directly verifies the related correction: Produce a reproducible versioned build artifact in the build stage and use that immutable artifact for subsequent validation and deployment. It checks the technical distinction: A CI build stage turns source and fixed dependencies into a traceable deployment input used by later stages. Applied here, the relevant facts are: An Ansible network workflow deploys source directly from a mutable branch without a versioned build artifact. The question asks what acceptance check would prove the intended behavior was restored.

Incorrect Answers

A:

The alternative verifies a different condition: This directly verifies the related correction: Run objective-based postdeployment checks and fail or roll back the change when observed network behavior contradicts intent. The current case instead depends on: This directly verifies the related correction: Produce a reproducible versioned build artifact in the build stage and use that immutable artifact for subsequent validation and deployment. Confirm that “The deployed artifact checksum matches the tested build artifact and the pipeline records its originating commit”.

B:

That test belongs to another control with this basis: This directly verifies the related correction: Resolve the conflicting lines deliberately against the approved desired network state, run validation, then commit the merged result. In the current case, the governing reason is: This directly verifies the related correction: Produce a reproducible versioned build artifact in the build stage and use that immutable artifact for subsequent validation and deployment. The direct verification is “The deployed artifact checksum matches the tested build artifact and the pipeline records its originating commit”.

C:

This observation would test a different technical dependency: This directly verifies the related correction: Provision the CML topology reproducibly and wait for all required node readiness conditions before tests begin. The current issue is governed by: This directly verifies the related correction: Produce a reproducible versioned build artifact in the build stage and use that immutable artifact for subsequent validation and deployment. Its decisive check is “The deployed artifact checksum matches the tested build artifact and the pipeline records its originating commit”.

Question 516

While investigating an inventory reconciliation for fictional Thornwick Health Trust, the team reports: An API orchestration stores refresh tokens in plain log output while attempting to recover from access-token expiration. What should the team add to its network automation operating process?

  1. Add timeout fault injection and per-target execution deadlines to the network automation test suite.
  2. Review handler notification conditions and test unchanged and changed execution paths before production scheduling.
  3. Make architecture reviews explicitly compare blast radius, recovery, security and long-term ownership.
  4. Test 401, refresh failure and 403 paths and enforce tightly scoped secret access and log redaction.

Explanation

Correct Answer

D:

Test 401, refresh failure and 403 paths and enforce tightly scoped secret access and log redaction. This prevents the recurrence because an access-token lifecycle and authorization scope are distinct concerns; repeated reuse of expired or insufficient tokens cannot resolve them. A matching immediate correction is: Use the supported token refresh flow with secret storage and distinguish unauthenticated 401 from authorized-but-forbidden 403 responses. Applied here, the relevant facts are: An API orchestration stores refresh tokens in plain log output while attempting to recover from access-token expiration. The question asks what should the team add to its network automation operating process.

Incorrect Answers

A:

The offered action targets: python HTTP clients do not guarantee a useful bounded response time unless the caller configures explicit timeouts. The case facts point instead to: an access-token lifecycle and authorization scope are distinct concerns; repeated reuse of expired or insufficient tokens cannot resolve them. Implement “Test 401, refresh failure and 403 paths and enforce tightly scoped secret access and log redaction” to address that specific condition.

B:

This alternative is a response to another technical cause: ansible handlers are triggered by changed tasks and coalesce notifications, separating configuration checks from disruptive service actions. The reported case instead depends on: an access-token lifecycle and authorization scope are distinct concerns; repeated reuse of expired or insufficient tokens cannot resolve them. The action that directly addresses it is “Test 401, refresh failure and 403 paths and enforce tightly scoped secret access and log redaction”.

C:

The proposed action belongs to the mechanism: automation choice is driven by lifecycle and operational risk rather than by a single preferred programming tool. Here the stated fault follows a different mechanism: an access-token lifecycle and authorization scope are distinct concerns; repeated reuse of expired or insufficient tokens cannot resolve them. The corrective choice is “Test 401, refresh failure and 403 paths and enforce tightly scoped secret access and log redaction”.

Question 517

At fictional Seabright Learning, an acceptance review of a reproducible simulation finds: A developer cherry-picks a small Terraform resource fix without its prerequisite helper update, and the maintenance pipeline cannot import a function. Which result most directly verifies that the proposed correction worked?

  1. The runner resolves the reviewed versions and integration tests pass consistently across repeated clean builds.
  2. A deliberately invalid network change blocks before deploy and a valid change reaches the deployment stage only after passing gates.
  3. Lab simulation reproduces the known routing dependency and the candidate change passes normal and failure scenarios.
  4. The target branch builds and tests cleanly with only the required backported changes and no missing dependency references.

Explanation

Correct Answer

D:

The target branch builds and tests cleanly with only the required backported changes and no missing dependency references. This directly verifies the related correction: Identify the prerequisite commits and cherry-pick a minimal compatible sequence or adapt the patch for the target baseline. It checks the technical distinction: A cherry-picked commit does not automatically bring every ancestor or dependent change from its original branch. Applied here, the relevant facts are: A developer cherry-picks a small Terraform resource fix without its prerequisite helper update, and the maintenance pipeline cannot import a function. The question asks which result most directly verifies that the proposed correction worked.

Incorrect Answers

A:

This alternative is a response to another technical cause: This directly verifies the related correction: Pin compatible build tool and package versions, update the lockfile and test upgrades in a separate controlled pipeline. The reported case instead depends on: This directly verifies the related correction: Identify the prerequisite commits and cherry-pick a minimal compatible sequence or adapt the patch for the target baseline. The action that directly addresses it is “The target branch builds and tests cleanly with only the required backported changes and no missing dependency references”.

B:

The proposed action belongs to the mechanism: This directly verifies the related correction: Make deployment depend on successful prevalidation jobs for schema, policy, simulation and approval checks. Here the stated fault follows a different mechanism: This directly verifies the related correction: Identify the prerequisite commits and cherry-pick a minimal compatible sequence or adapt the patch for the target baseline. The corrective choice is “The target branch builds and tests cleanly with only the required backported changes and no missing dependency references”.

C:

The offered action targets: This directly verifies the related correction: Model the relevant links, routing peers and failure cases in CML rather than only a minimal happy-path device. The case facts point instead to: This directly verifies the related correction: Identify the prerequisite commits and cherry-pick a minimal compatible sequence or adapt the patch for the target baseline. Implement “The target branch builds and tests cleanly with only the required backported changes and no missing dependency references” to address that specific condition.

Question 518

At fictional Northshore Cloud, an acceptance review of an inventory reconciliation finds: A team selects a long bespoke Python script for static VLAN declarations that a supported idempotent provider can already manage reliably. Which targeted correction best addresses the network automation failure?

  1. Separate inventory groups and group variables by approved device scope, then constrain the play to the intended group.
  2. Respect rate-limit response headers and coordinate bounded exponential backoff with jitter under the shared API budget.
  3. Move credentials into a protected secret store or Ansible Vault and suppress sensitive task output while keeping access narrowly scoped.
  4. Choose a supported declarative IaC framework for persistent desired configuration and targeted scripts for bounded imperative queries or integrations.

Explanation

Correct Answer

D:

Choose a supported declarative IaC framework for persistent desired configuration and targeted scripts for bounded imperative queries or integrations. This addresses the problem because infrastructure as code, low-code workflows and custom software solve different state-lifecycle and integration needs. Confirm success using: The design review shows the chosen approach fits the resource lifecycle, state ownership and change frequency without redundant infrastructure code.

Incorrect Answers

A:

The proposed action belongs to the mechanism: ansible inventory groups and variable precedence determine which devices and values are targeted by a playbook. Here the stated fault follows a different mechanism: infrastructure as code, low-code workflows and custom software solve different state-lifecycle and integration needs. The corrective choice is “Choose a supported declarative IaC framework for persistent desired configuration and targeted scripts for bounded imperative queries or integrations”.

B:

The offered action targets: hTTP 429 signals server-enforced rate limits; API client backoff must consider shared rate budgets and Retry-After semantics. The case facts point instead to: infrastructure as code, low-code workflows and custom software solve different state-lifecycle and integration needs. Implement “Choose a supported declarative IaC framework for persistent desired configuration and targeted scripts for bounded imperative queries or integrations” to address that specific condition.

C:

This alternative is a response to another technical cause: automation credentials need confidentiality at rest and in output; inventory convenience does not justify committing secrets. The reported case instead depends on: infrastructure as code, low-code workflows and custom software solve different state-lifecycle and integration needs. The action that directly addresses it is “Choose a supported declarative IaC framework for persistent desired configuration and targeted scripts for bounded imperative queries or integrations”.

Question 519

During a telemetry quality investigation at fictional Foxglove Aviation, the change record states: A Python script intermittently fails after a fixed request timeout; a retry count is known but the upstream response duration is not recorded. Which implementation action best meets the observed technical requirement?

  1. Generate and store private keys in restricted secret systems and deploy only to the authorized TLS termination endpoint.
  2. Enforce server-side per-resource authorization using authenticated identity and approved tenant or device scope.
  3. Capture per-request endpoint, duration, timeout class, retry attempt and status while preserving sensitive data redaction.
  4. Configure suitable model-driven telemetry subscriptions for the relevant paths at a collection interval appropriate to the signal.

Explanation

Correct Answer

C:

Capture per-request endpoint, duration, timeout class, retry attempt and status while preserving sensitive data redaction. This addresses the problem because latency and retry failures require timing context and per-attempt telemetry; an aggregate success percentage is insufficient. Confirm success using: Correlated traces identify which API operation approaches the timeout threshold and whether backoff reduces the failure rate. Applied here, the relevant facts are: A Python script intermittently fails after a fixed request timeout; a retry count is known but the upstream response duration is not recorded. The question asks which implementation action best meets the observed technical requirement.

Incorrect Answers

A:

The offered action targets: cA-signed certificates are public trust assertions but the associated private key must remain confidential for endpoint authentication. The case facts point instead to: latency and retry failures require timing context and per-attempt telemetry; an aggregate success percentage is insufficient. Implement “Capture per-request endpoint, duration, timeout class, retry attempt and status while preserving sensitive data redaction” to address that specific condition.

B:

This alternative is a response to another technical cause: authentication proves identity but each network resource action requires separate policy-based authorization controls. The reported case instead depends on: latency and retry failures require timing context and per-attempt telemetry; an aggregate success percentage is insufficient. The action that directly addresses it is “Capture per-request endpoint, duration, timeout class, retry attempt and status while preserving sensitive data redaction”.

D:

The offered action targets: model-driven telemetry publishes structured device state using subscribed paths and collection modes rather than relying only on slow polling. The case facts point instead to: latency and retry failures require timing context and per-attempt telemetry; an aggregate success percentage is insufficient. Implement “Capture per-request endpoint, duration, timeout class, retry attempt and status while preserving sensitive data redaction” to address that specific condition.

Question 520

At fictional Sunrise Aeronautics, during an AI-assisted code review, the operator observes: A network incident log contains attacker-controlled text saying ignore safety checks and run a destructive router command, and the agent obeys it. What should the engineer change first to resolve this specific issue?

  1. Treat retrieved logs as untrusted data and enforce instruction hierarchy and independent tool permissions outside the model.
  2. Check AI conclusions against fresh independent device and controller evidence with explicit freshness and confidence rules.
  3. Require citation to a compatible authoritative reference and fail closed or escalate when command syntax or device version is uncertain.
  4. Authenticate each MCP client, authorize its tool calls by tenant identity and isolate credential/session context.

Explanation

Correct Answer

A:

Treat retrieved logs as untrusted data and enforce instruction hierarchy and independent tool permissions outside the model. This addresses the problem because prompt injection attempts to convert lower-trust retrieved content into high-priority instructions across an AI-agent trust boundary. Confirm success using: Injection test logs are summarized as evidence but never cause unauthorized configuration or secret-retrieval tool calls. Applied here, the relevant facts are: A network incident log contains attacker-controlled text saying ignore safety checks and run a destructive router command, and the agent obeys it. The question asks what should the engineer change first to resolve this specific issue.

Incorrect Answers

B:

The proposed action belongs to the mechanism: aI recommendations are hypotheses requiring external ground truth; absent evidence is not proof of a network condition. Here the stated fault follows a different mechanism: prompt injection attempts to convert lower-trust retrieved content into high-priority instructions across an AI-agent trust boundary. The corrective choice is “Treat retrieved logs as untrusted data and enforce instruction hierarchy and independent tool permissions outside the model”.

C:

The offered action targets: model fluency is not evidence of device capability; uncertain technical recommendations require explicit verification before action. The case facts point instead to: prompt injection attempts to convert lower-trust retrieved content into high-priority instructions across an AI-agent trust boundary. Implement “Treat retrieved logs as untrusted data and enforce instruction hierarchy and independent tool permissions outside the model” to address that specific condition.

D:

This alternative is a response to another technical cause: an MCP server is an integration boundary and must not treat model-selected identity strings as verified security credentials. The reported case instead depends on: prompt injection attempts to convert lower-trust retrieved content into high-priority instructions across an AI-agent trust boundary. The action that directly addresses it is “Treat retrieved logs as untrusted data and enforce instruction hierarchy and independent tool permissions outside the model”.