AI Agents Tutorial menu

AI Agents Quiz

Check your understanding of AI agents with this quiz covering the whole tutorial.

How the Quiz Works

There are ten questions. Choose an answer for each one to see if it is correct, along with a short explanation. If you get one wrong, go back to that lesson and read it again.

Questions	Topics
1 to 3	What agents are, tools and function calling
4 to 6	The agent loop, memory and multi-agent systems
7 to 10	Building an agent, risks and human approval

Questions

Exercise

1. Which task is the best fit for an AI agent rather than a fixed automation?

Sending the same report email every night at 9 pmFinding out why sales dropped last month and drafting a short reportTurning on a light at sunsetBacking up a folder every Sunday

Exercise

2. When a model makes a tool call, what does it actually send back?

The result of running the functionThe tool's name and the arguments it wants, for your code to runThe Python source code of the toolNothing, the tool runs silently

Exercise

3. Why is a clear tool description so important?

It makes the tool run fasterIt is shown to the user as the replyThe model uses it to decide when and how to use the toolIt encrypts the arguments

Exercise

4. What are the steps of the basic agent loop?

Download, install, restartThink, act with a tool, observe the result, then repeat or finishAsk the user a question at every stepWrite the final answer first, then call tools

Exercise

5. Does a language model remember earlier tasks by itself between API requests?

Yes, it learns from every chat automaticallyNo, your code must store memory and send the relevant parts back in the promptOnly if the user types slowlyYes, but only for one week

Exercise

6. When should you split a job between several agents?

Always, more agents are always betterWhen one agent with good tools clearly struggles, for example because the task is too large or needs separate skillsOnly when the task is very smallNever, agents cannot work together

Exercise

7. Why test an agent loop with a fake model first?

Fake models are smarter than real onesYou can check your loop and tools for free, with repeatable results, before calling a real APIReal APIs do not support toolsIt makes the agent run in the browser

Exercise

8. What is prompt injection?

Typing a prompt very fastAdding more examples to a promptHidden instructions in content the agent reads, such as a web page or email, that try to change what it doesA way to make answers shorter

Exercise

9. Which action should normally wait for human approval?

Reading an order's delivery statusSearching the shop's price listIssuing a Rs 25,000 refund to a customerAdding two numbers with a calculator tool

Exercise

10. Where should step and spending limits for an agent be enforced?

Only in the system promptIn your code, with a backup spending limit in the AI provider's dashboardNowhere, agents stop on their ownIn the user's browser history

What Next?

Scored 8 or more? Build your own small agent for a real task, like sorting your emails or summarising a sales file, and keep a human in the loop. Scored less? Revisit the lessons on tools, the agent loop and risks, then try again.

Learn AI Agents with a teacher

Join the AI & Prompt Engineering course at Workownix, online or in Orangi Town, Karachi, and get a certificate.

See the course Ask on WhatsApp