Skip to content
AI Security

Interview question · Agent, Tool & MCP Security

How would you protect AI agents at runtime: sandboxes, syscall monitoring, tool-call policy and data provenance?

Design-time controls decide what an agent may do. Runtime protection watches what it actually does, at four different layers, and each layer catches something the others miss.

expertNewPreparation tags: Google DeepMind, Microsoft, AmazonUpdated Nov 2026

Design-time controls decide what an agent may do. Runtime protection watches what it actually does, at four different layers, and each layer catches something the others miss.

Make your first attempt

What would you say before reading the answer?

  1. 01 State the assumptions your answer depends on.
  2. 02 Explain the mechanism and choose a control.
  3. 03 Name a condition that would change your decision.

The worked answer includes an approach, scoring criteria and follow-up questions. Compare the reasoning with your own.

10 answers per topic instead of 5, plus saved progress and bookmarks · costs nothingor unlock all 529 remaining answers · $25

Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, the complete practice-test bank, and the downloadable security field guides.

6 months · One payment · No auto-renewal

What if · the follow-ups in this answer

The answer works through 5 follow-ups like an interviewer would

A research agent reads web pages, writes and runs Python to analyse data, and can email reports. A security review asks what protects the company while it runs, not just at design time.

  1. Where should the agent's Python run?
  2. What would you watch inside the sandbox?
  3. A web page tells the agent to email its findings to an outside address. What stops it?
  4. What happens when one of these layers fires?
  5. What is the runtime protection design?

Each rung has a nudge, a model answer, the principle that carries over and the weak answer to avoid. Premium opens every answer and ladder.

UP NEXT ON YOUR JOURNEY
Next in this trackYour agent will call other companies' agents over the A2A protocol. What are the security risks, and what does a signed Agent Card prove?Next in this trackWhen an AI agent buys things for users, how do payment protocols prove consent, cap spending and let merchants trust the agent?Next in this trackHow is an agent skill from a public marketplace different from an MCP server as a risk, and how would you govern skills?
DISCUSSION · 0

No comments yet. Share a correction or a different way to answer it.