Prompt injection defense in Prompt Engineering / GenAI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to safely add user input to the prompt.

Prompt Engineering / GenAI

safe_prompt = base_prompt + ' User says: ' + [1]

Drag options to blanks, or click blank then click option'

Auser_input

Bsanitize(user_input)

Craw_input

Dinput()

Attempts:

3 left

💡 Hint

Common Mistakes

Using raw user input without cleaning.

Using input() which reads new input instead of the variable.

Using sanitize(user_input) helps remove harmful parts from the input before adding it to the prompt.

2fill in blank

medium

Complete the code to detect prompt injection keywords.

Prompt Engineering / GenAI

if '[1]' in user_input.lower():
    alert('Possible injection detected')

Drag options to blanks, or click blank then click option'

Aignore previous command

Bdelete all

Cignore previous

Dignore previous instructions

Attempts:

3 left

💡 Hint

Common Mistakes

Using incomplete or incorrect phrases that don't match injection attempts.

Checking for unrelated words.

The phrase 'ignore previous instructions' is a common injection attempt to override the prompt.

3fill in blank

hard

Fix the error in the function that blocks injection by replacing dangerous words.

Prompt Engineering / GenAI

def block_injection(text):
    blocked_text = text.replace([1], '[REDACTED]')
    return blocked_text

Drag options to blanks, or click blank then click option'

A'ignore previous instructions'

Bignore previous instructions

C'ignore previous'

D'delete all'

Attempts:

3 left

💡 Hint

Common Mistakes

Passing a variable name without quotes causes a NameError.

Using incomplete phrases that don't match the injection.

The replace method needs the first argument as a string literal, so it must be in quotes.

4fill in blank

hard

Fill both blanks to create a safe prompt by filtering and then formatting user input.

Prompt Engineering / GenAI

def create_safe_prompt(user_input):
    filtered = [1](user_input)
    prompt = f"System: Follow rules. User says: [2]"
    return prompt

Drag options to blanks, or click blank then click option'

Asanitize

Buser_input

Cfiltered

Dclean_input

Attempts:

3 left

💡 Hint

Common Mistakes

Using the original user_input directly in the prompt.

Mixing variable names incorrectly.

First sanitize the input, then use the filtered result in the prompt string.

5fill in blank

hard

Fill all three blanks to check for injection and respond safely.

Prompt Engineering / GenAI

def respond(user_input):
    if [1] in user_input.lower():
        return '[2]'
    safe_input = [3](user_input)
    return f"Processed: {safe_input}"

Drag options to blanks, or click blank then click option'

A'ignore previous instructions'

B'Injection detected. Input blocked.'

Csanitize

Duser_input

Attempts:

3 left

💡 Hint

Common Mistakes

Not checking for the exact injection phrase.

Returning unsafe input directly.

Not sanitizing input before use.

Check if the injection phrase is in input, return a warning, else sanitize input before processing.