📖 Lesson content
Your score:
23 of 23 Correct (100%)
Elapsed time:
21 minutes
Show Answers
Hide Answers
Question 1: Correct answer
What is a tool function in the context of Claude's tool use system?
A special API endpoint provided by Anthropic
A configuration file that defines Claude's behavior
A database query that retrieves user preferences
A plain function that gets executed when Claude needs additional information or needs to perform an action
Question 2: Correct answer
You're improving a prompt that isn't working well. What should you do after applying a prompt engineering technique?
Start over with a completely new prompt
Ask someone else to write the prompt
Use prompt evaluations to see if it actually improved
Move on to the next technique immediately
Question 3: Correct answer
You're making a math tutoring app. You want Claude to give hints instead of direct answers. What should you use?
Ask users to say "please" in every question
Use smaller fonts to make Claude quiet
Send messages only on weekdays
A system prompt explaining Claude should act like a tutor
Question 4: Correct answer
You want Claude to write very creative, unpredictable stories. What temperature setting should you use?
0.5 (medium)
0.0 (very low)
Temperature doesn't affect creativity
1.0 (very high)
Question 5: Correct answer
You want an AI to write a product description. Which opening is most clear and direct?
"I was wondering if you could maybe help me with something about products?"
"Can you tell me about products and descriptions and stuff?"
"Write a product description for running shoes."
"What do you think makes a good product description?"
Question 6: Correct answer
You want to measure how well your AI prompt actually works in practice. Which approach should you focus on?
Using multishot prompting examples
Prompt engineering techniques like XML tags
Writing longer, more detailed prompts
Prompt evaluation with automated testing
Question 7: Correct answer
You're building a web app that talks to Claude. Where should you store your API key?
On your server, hidden from users
In a public GitHub repository
In the web browser's settings
In your website's JavaScript code
Question 8: Correct answer
What is a model grader in prompt evaluation?
A human reviewer who manually scores AI outputs
Another AI model used to assess the quality of outputs
A programmatic check that validates syntax and format
A scoring system that only measures response speed
Question 9: Correct answer
You're asking an AI to analyze both customer reviews and sales data. How should you organize this information in your prompt?
Put reviews first, then sales data with no separators
Mix everything together in one paragraph
Write the data in different fonts
Use XML tags like and <sales_data> to separate them
Question 10: Correct answer
You want to send a message to Claude through the API. Which four things do you absolutely need to include?
Temperature, creativity, speed, and language
Date, time, location, and device type
API key, model name, messages, and max tokens
Username, password, email, and phone number
Question 11: Correct answer
What is the primary purpose of tool use in Claude?
To automatically save conversation history
To allow Claude to access real-time information and external systems beyond its training data
To make Claude respond faster to user queries
To reduce the number of tokens Claude uses in responses
Question 12: Correct answer
You're running a prompt evaluation. After getting responses from Claude, what's the next step in the typical workflow?
Rewrite the prompt completely from scratch
Create a new dataset with different questions
Deploy the prompt to production immediately
Feed the responses through a grader for scoring
Question 13: Correct answer
You ask Claude "What is pizza?" and it answers. Then you ask "What toppings are popular?" but Claude doesn't know you're still talking about pizza. What's the problem?
You need to send the whole conversation history with each request
Claude doesn't like pizza
Claude is broken
You're asking too many questions
Question 14: Correct answer
What is the primary difference between an MCP Server and an MCP Client in terms of their roles?
MCP Servers contain tools, prompts, and resources while MCP Clients act as the communication bridge to access those tools
MCP Servers handle user authentication while MCP Clients manage permissions
MCP Servers store data while MCP Clients process requests
MCP Servers run on remote machines while MCP Clients only run locally
Question 15: Correct answer
Which of the following best describes Computer Use in the context of Claude?
A method for deploying Claude applications to cloud servers
A billing system that tracks how much computational time Claude consumes
A capability that lets Claude interact directly with desktop environments like a human would
A feature that optimizes Claude's processing speed on different hardware
Question 16: Correct answer
What does "transport agnostic" mean in the context of MCP communication?
MCP clients and servers can communicate using different methods like HTTP or standard input/output
MCP only works with encrypted connections
MCP automatically chooses the fastest available network connection
MCP requires specific hardware to function properly
Question 17: Correct answer
What is the primary purpose of a batch tool in Claude's tool system?
To accept multiple tool calls and execute them simultaneously
To encrypt multiple tool calls for security purposes
To process large files that exceed normal size limits
To automatically retry failed tool calls
Question 18: Correct answer
Claude responds to your request with both explanatory text and a tool use block. What type of message structure is this?
An error message that needs to be fixed
A multi-block message with different content types
A single-block text message
A malformed response that should be ignored
Question 19: Correct answer
You're building an app where users need to verify information Claude provides from documents. What feature should you enable?
Code execution
Prompt caching
Citations
Extended thinking
Question 20: Correct answer
When should you choose workflows over agents for handling user tasks?
When you're not sure what tasks or parameters you'll give to Claude
When you need maximum flexibility in task completion
When you want Claude to creatively combine tools in unexpected ways
When you can picture the exact flow or steps Claude should go through to solve a problem
Question 21: Correct answer
Which of the following best describes why environment inspection is crucial for AI agents?
It helps agents run faster and more efficiently
It allows agents to observe and understand the results of their actions
It simplifies the agent's tool requirements
It reduces the cost of running AI operations
Question 22: Correct answer
What is the Model Context Protocol (MCP)?
A security protocol for protecting AI model parameters
A communication layer that provides Claude with context and tools without requiring tedious integration code
A programming language designed specifically for AI applications
A database system for storing AI training data
Question 23: Correct answer
You keep sending the same long document to Claude with different questions. How can you make this faster and cheaper?
Compress the document first
Split the document into smaller pieces
Use prompt caching with cache breakpoints
Ask multiple questions at once
Take this again
🔁 Related lessons
- Next: Course Wrap Up
- Previous: Quiz on Agents and Workflows
- Same section: Making a request · Multi-Turn conversations · Chat exercise
- Part of paths: Path C
- Reference docs: Glossary · Skills atlas · By use-case
📚 Source & attribution
- Original Anthropic Academy lesson: https://anthropic.skilljar.com/claude-with-the-anthropic-api/290899
- © 2025 Anthropic. Educational fair-use only.