CCNA v2.0 Practice Questions: AI, Network Operations and Management

Correct answer: B. The agent gathered evidence with read-only commands, traced the errors to a duplex mismatch and held the change for approval

B is correct because heavy CRC errors on R1 Gi0/2 and half duplex on the neighbor port support a duplex mismatch, and the change waits for a human. A misdescribes agentic AI, which is defined by calling tools such as show commands to gather data. C ignores the guardrail that configuration changes need human approval, even when the evidence looks clear. D would guess at a fix before collecting evidence, and setting half duplex would not be the right correction anyway.

Correct answer: A. Verify that 10.0.12.1 is a reachable neighbor that knows the 10.50.1.0/24 network before approving the static route

A is correct because the missing route is supported by evidence, but the agent itself flags the next hop as unverified, so it must be checked first. B is too broad; recommending a change for human review is exactly how a guarded agent is meant to work. C widens the change far beyond the evidence and the one router that is missing the route. D ignores the agent's own note that the next hop was never verified with any command.

Correct answer: B. The conclusion is not supported; the administratively down state shows Gi0/1 was shut down, and the log was never checked

B is correct because administratively down means a shutdown command, not a cabling fault, and the agent acted on incomplete data after its log check timed out. A misreads the output; up up means Gi0/0 is working, and the problem interface is Gi0/1. C ignores the administratively down status, which means the interface was shut down regardless of the cable. D treats the ticket history as proof and ignores the interface state, which points to configuration rather than the cable.

Correct answer: D. It exposes a pre-shared key and customer details to a public model

D is correct because the IKE pre-shared key, customer name and real public addresses are sensitive data that should be redacted or sent only to an approved internal model. A misreads the prompt, which asks only for a list of likely causes. B is false; the prompt asks for a numbered list aimed at a CCNA-level technician. C is false; the prompt opens with a senior network engineer persona.

Correct answer: D. Split the request so the prompt asks one task: identify the cause of the VLAN 10 loop from the output

D is correct because the prompt bundles four unrelated tasks, so focusing on one task with clear output gets a better and more checkable answer. A is wrong; the model cannot know the platform, and the version helps it give correct syntax. B has no basis and would move internal data to a less controlled model without any benefit. C is wrong; relevant show output gives the model facts to reason from and reduces guessing.

Correct answer: D. Replace the community string with a placeholder such as COMMUNITY and keep the rest of the prompt as written

D is correct because the community string is a credential that must not reach a public model, while the persona, platform, format and task are already sound. A adds more credentials to a prompt for a public model, which makes the data exposure worse. B takes away context that steers the answer toward correct IOS XE syntax at the right level. C removes a useful component; a clear format helps the technician follow and check the answer.

Correct answer: D. It reports VLAN 40 as configuration drift and, with state overridden, removes it so SW1 matches the file

D is correct because the repository is the source of truth, and state overridden makes the device VLAN list match the file exactly, so the unlisted VLAN 40 is drift and is removed. A has the outcome backwards; the file defines VLANs 10, 20 and 30 as the desired state. B reverses the IaC model; the files are the source of truth and the device is made to match them. C is false; the desired state lists VLANs explicitly, so the tool compares them.

Correct answer: B. Revert commit 9b07d11 through a reviewed change and let the pipeline apply the restored desired state

B is correct because 9b07d11 is the commit merged yesterday, and reverting it restores the previous desired state in Git, and the pipeline then pushes it, keeping history and review intact. A destroys the change history and turns the faulty device state into the new source of truth. C creates drift, because the repository still holds the bad state and the pipeline would reapply it. D targets the wrong commit; removing the baseline would undo the whole building configuration.

Correct answer: B. Idempotence, because applying the same desired state again makes no further changes

B is correct because the devices already match the desired state after the first run, so the second run changes nothing, which is idempotent behavior. A would show changed counts, and nothing in the files asked the tool to undo anything. C is the opposite reading; changed=0 shows the switches already match the desired state. D describes high availability, which has nothing to do with the repeated run results.

Correct answer: B. Anyone who captures FGwrite can send Set requests that change R1

B is correct because SNMPv2c sends community strings in clear text, and the RW community allows Set operations that modify the device. A is wrong because SNMPv2c never encrypts community strings, and a location string is not the main risk. C is false because community strings alone are enough for an SNMPv1 or v2c manager to poll the agent. D reverses the meaning; an RO community permits only Get, GetNext and GetBulk requests.