AI in Network Operations: practice questions

Predictive, generative and agentic AI in network operations, and writing a prompt for a network task.

CCNA 200-301 v2.0, exam topic 5.1-5.2 · part of 5. AI, and Network Operations and Management, 10% of the exam

The same material in the CCNA 200-301 v1.1 exam: Automation Impact

What this topic covers

Predictive and generative AI
Predictive models flag anomalies and forecast load from telemetry; generative models draft text, configurations and summaries.
Agentic AI
An agent plans and carries out multi-step tasks with tools, so it needs limited permissions and human approval for changes.
Writing a prompt
A useful prompt sets a persona, gives instructions and context, asks for an output format and leaves out confidential data.

Sample question

Try it, then open the answer. Our questions are written by AI and checked independently before they are published.

Easy Single choice

Refer to the exhibit. An agentic AI workflow investigating an OSPF neighbor issue stops unexpectedly. What is the cause of the failure?

Execution Trace Log
[Step 1] Agent -> Call Tool: run_cli_command(device='R1', command='show interfaces GigabitEthernet0/1')
[Step 1] System -> Response: "GigabitEthernet0/1 is up, line protocol is up, MTU 1500 bytes ..."
[Step 2] Agent -> Call Tool: run_cli_command(device='R2', command='show interfaces GigabitEthernet0/1')
[Step 2] System -> Response: "GigabitEthernet0/1 is up, line protocol is up, MTU 9000 bytes ..."
[Step 3] Agent -> Call Tool: apply_device_config(device='R1', interface='GigabitEthernet0/1', config='mtu 9000')
[Step 3] System -> Guardrail: BLOCKED - POLICY_VIOLATION: Mutation action requested without valid human_approval_token
[Step 4] Agent -> System: Execution halted with error status code 403
  1. A. The agent attempted a configuration change without an approval checkpoint.
  2. B. The agent passed an invalid argument to the configuration tool.
  3. C. The agent was restricted from executing diagnostic CLI commands.
  4. D. The agent exceeded the maximum allowable diagnostic query retries.
Show the answer

Answer: A. The agent attempted a configuration change without an approval checkpoint.

Agentic AI systems use reasoning and tool invocations to autonomously collect operational telemetry and form remediation plans. However, to maintain network safety and prevent accidental disruption, enterprise guardrails enforce limited permissions and require an explicit human approval checkpoint before executing state-changing (mutation) commands. The trace shows the agent diagnosing the MTU mismatch via read-only tools and then attempting to apply 'mtu 9000' directly, which the guardrail blocked due to the missing approval token.

Further reading