Semantic Bug Detector

Semantic Bug Detector

Detect bugs where code behavior doesn't match its intended purpose.

Overview

This skill analyzes code to find semantic bugs—errors where the implementation contradicts the intent expressed through names, comments, and documentation. Unlike syntax errors or type errors, semantic bugs are logically valid code that does the wrong thing.

How to Use

Provide code with any of:

1. Function/variable names that express intent
2. Comments describing what code should do
3. Docstrings specifying behavior
4. Documentation stating requirements

The skill will:

• Infer intended behavior from these sources
• Analyze actual implementation
• Identify mismatches
• Report semantic bugs with explanations

Detection Workflow

Step 1: Extract Intent

Gather intent signals from multiple sources:

Names: is_even, get_last_n_elements, calculate_average

• Infer expected behavior from naming conventions
• Identify predicates (is_, has_, can_)
• Recognize operations (get_, set_, calculate_)

Comments: // Returns first n elements, # Check if x is positive

• Parse inline comments
• Extract stated purpose
• Identify boundary specifications

Docstrings:

"""Calculate the average of a list of numbers.
Returns the sum divided by the count."""

• Parse structured documentation
• Extract preconditions and postconditions
• Identify range specifications

Step 2: Analyze Implementation

Examine actual code behavior:

Control flow: Conditions, loops, branches Operations: Arithmetic, logical, comparison operators Boundaries: Array indices, range limits Edge cases: Empty input, null values, zero divisors

Step 3: Compare Intent vs Implementation

Check for common mismatches:

Off-by-one errors: Using n+1 when should use n Inverted logic: Returning opposite boolean value Wrong operator: Using * when should use / Boundary errors: Inclusive when should be exclusive Missing checks: Not handling empty/null input

Step 4: Report Findings

For each bug found, provide:

• Location: Function/line where bug occurs
• Intent: What the code should do
• Actual: What the code actually does
• Bug type: Category of semantic error
• Fix: Suggested correction

Example: Off-by-One Error

Code:

def get_last_n_elements(arr, n):
    """Returns the last n elements from the array."""
    return arr[-n-1:]

Analysis:

1. Intent from name: "get_last_n_elements" → should return exactly n elements
2. Intent from docstring: "Returns the last n elements" → confirms n elements
3. Actual behavior: arr[-n-1:] returns n+1 elements
4. Mismatch: Returns n+1 instead of n

Report:

BUG: Off-by-one error in get_last_n_elements

Location: Line 3, return statement
Intent: Return the last n elements (from name and docstring)
Actual: Returns the last n+1 elements
Bug Type: Off-by-one error
Severity: High

Explanation:
The slice arr[-n-1:] starts at index -(n+1), which includes
one extra element. Should use arr[-n:] to get exactly n elements.

Fix:
return arr[-n:]

Example: Inverted Logic

Code:

def is_even(x):
    """Check if x is even."""
    return x % 2 == 1

Analysis:

1. Intent from name: "is_even" → should return True for even numbers
2. Intent from docstring: "Check if x is even" → confirms even check
3. Actual behavior: x % 2 == 1 returns True for odd numbers
4. Mismatch: Logic is inverted

Report:

BUG: Inverted logic in is_even

Location: Line 3, return statement
Intent: Return True when x is even (from name and docstring)
Actual: Returns True when x is odd
Bug Type: Inverted boolean logic
Severity: High

Explanation:
x % 2 == 1 is True for odd numbers, not even numbers.
The condition is inverted from the stated intent.

Fix:
return x % 2 == 0

Example: Boundary Mismatch

Code:

def in_range(x, start, end):
    """Check if x is in range [start, end)."""
    return start <= x <= end

Analysis:

1. Intent from docstring: "[start, end)" → half-open interval, excludes end
2. Actual behavior: start <= x <= end includes end
3. Mismatch: Uses inclusive end when should be exclusive

Report:

BUG: Boundary mismatch in in_range

Location: Line 3, return statement
Intent: Check if x in [start, end) - half-open interval (from docstring)
Actual: Checks if x in [start, end] - closed interval
Bug Type: Boundary error (inclusive vs exclusive)
Severity: Medium

Explanation:
The notation [start, end) means start is included but end is excluded.
The condition start <= x <= end includes end, violating the spec.

Fix:
return start <= x < end

Example: Wrong Operator

Code:

def calculate_average(numbers):
    """Calculate the average of a list of numbers."""
    return sum(numbers) * len(numbers)

Analysis:

1. Intent from name: "calculate_average" → should compute mean
2. Intent from docstring: "Calculate the average" → confirms mean calculation
3. Actual behavior: Multiplies sum by count instead of dividing
4. Mismatch: Wrong arithmetic operator

Report:

BUG: Wrong operator in calculate_average

Location: Line 3, return statement
Intent: Calculate average (sum / count) from name and docstring
Actual: Calculates sum * count
Bug Type: Wrong arithmetic operator
Severity: High

Explanation:
Average is calculated by dividing sum by count, not multiplying.
Using * instead of / produces incorrect result.

Fix:
return sum(numbers) / len(numbers)

Example: Missing Edge Case

Code:

def find_max(numbers):
    """Find the maximum number in the list."""
    max_val = numbers[0]
    for num in numbers[1:]:
        if num > max_val:
            max_val = num
    return max_val

Analysis:

1. Intent from name: "find_max" → should find maximum
2. Intent from docstring: "Find the maximum number in the list"
3. Actual behavior: Crashes on empty list (IndexError)
4. Mismatch: Doesn't handle empty input

Report:

BUG: Missing edge case handling in find_max

Location: Line 3, accessing numbers[0]
Intent: Find maximum number in list (from name and docstring)
Actual: Crashes with IndexError when list is empty
Bug Type: Missing edge case (empty input)
Severity: High

Explanation:
The function assumes the list is non-empty by accessing numbers[0]
without checking. This causes a crash on empty input.

Fix:
if not numbers:
    raise ValueError("Cannot find max of empty list")
max_val = numbers[0]
...

Common Bug Categories

Off-by-One Errors

Indicators: "first n", "last n", "range", "iterate" Bugs: Using n+1 instead of n, <= instead of < See: bug_patterns.md

Inverted Logic

Indicators: "is_", "has_", "can_", boolean predicates Bugs: Returning opposite value, wrong comparison See: bug_patterns.md

Boundary Mismatches

Indicators: "[a, b]", "[a, b)", range specifications Bugs: Inclusive when should be exclusive See: bug_patterns.md