Output filtering and safety checks in Agentic AI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to add a simple output filter that blocks offensive words.

Agentic AI

def filter_output(text):
    blocked_words = ['badword1', 'badword2']
    for word in blocked_words:
        if word in text:
            return [1]
    return text

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

A""

B"[Content blocked]"

Ctext

DNone

Attempts:

3 left

💡 Hint

Common Mistakes

Returning the original text even when blocked words are found.

Returning None which might cause errors downstream.

The function returns a placeholder string when blocked words are found to prevent offensive output.

2fill in blank

medium

Complete the code to check if the model output length exceeds the safety limit.

Agentic AI

def check_length(output):
    max_length = 100
    if len(output) [1] max_length:
        return False
    return True

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

A!=

B<

C==

D>

Attempts:

3 left

💡 Hint

Common Mistakes

Using '<' which would block short outputs instead.

Using '==' which only blocks outputs exactly equal to max_length.

The function returns False if the output length is greater than the max allowed length.

3fill in blank

hard

Fix the error in the code that filters outputs containing sensitive keywords.

Agentic AI

def safe_output(text):
    sensitive_keywords = ['password', 'secret']
    if any([1] in text for [2] in sensitive_keywords):
        return '[Filtered]'
    return text

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

Ak, keyword

Bkeyword, k

Ckeyword, keyword

Dk, k

Attempts:

3 left

💡 Hint

Common Mistakes

Swapping the variable names causing syntax errors.

Using the same variable name twice causing confusion.

The correct syntax is to check if 'k' (each keyword) is in text, iterating over 'keyword' in sensitive_keywords.

4fill in blank

hard

Fill both blanks to create a dictionary filtering outputs by length and keyword presence.

Agentic AI

outputs = ['safe text', 'too long text example', 'contains secret']
filtered = {text: len(text) for text in outputs if len(text) [1] 15 and 'secret' not in text [2]

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

A<

B>

Cand

Dor

Attempts:

3 left

💡 Hint

Common Mistakes

Using '>' which would filter out short texts.

Using 'or' which would allow texts with 'secret'.

We want texts shorter than 15 and that do not contain 'secret', so use '<' and 'and'.

5fill in blank

hard

Fill all three blanks to implement a safety check that blocks outputs with banned words or too long length.

Agentic AI

def safety_check(output):
    banned_words = ['hack', 'attack']
    max_len = 50
    if any([1] in output for [2] in banned_words) or len(output) [3] max_len:
        return False
    return True

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

Aword

Bw

C>

D<

Attempts:

3 left

💡 Hint

Common Mistakes

Using '<' for length which would block short outputs.

Using inconsistent variable names in the comprehension.

We check if any banned word is in output using 'word' and 'w' as variables, and block if length is greater than max_len.