Detect AI-generated anti-patterns: magic numbers, comment-over-code, over-engineering

[rudra496]

Summary

AI-generated code has distinctive anti-patterns that differ from human-written code. Add a new AIPatternsAnalyzer that detects these.

AI Code Anti-Patterns to Detect

1. Magic Numbers

AI loves magic numbers instead of named constants:

# AI-generated
if confidence > 0.85:
    retry(max_attempts=3, delay=2.5)

# Human-written
HIGH_CONFIDENCE_THRESHOLD = 0.85
MAX_RETRIES = 3
RETRY_DELAY_SECONDS = 2.5

2. Comment-Over-Code Ratio

AI tends to over-document obvious code:

# Initialize an empty list to store the results
results = []

# Loop through each item in the items list
for item in items:
    # Append the item to the results list
    results.append(item)

3. Over-Engineering

AI wraps simple logic in unnecessary abstractions:

class DataProcessorFactory:
    @staticmethod
    def create_processor(type: str) -> DataProcessor:
        # For a simple list filter...

4. Uniform Variable Names

AI tends to use result, data, item, value without descriptive names.

5. TODO/Placeholder Comments

AI leaves # TODO: implement this, # Add more error handling here, pass

Implementation

1. Create ai_trust_validator/analyzers/ai_patterns.py
2. Implement AIPatternsAnalyzer(BaseAnalyzer)
3. Add heuristics with configurable thresholds
4. Add to Validator._init_analyzers()
5. Add comprehensive tests

Acceptance Criteria

• Detects at least 5 AI-specific anti-patterns
• Comment-over-code ratio check works
• Magic number detection with configurable threshold
• Results contribute to trust score
• Tests with real AI-generated code samples

Difficulty

Intermediate — pattern matching is straightforward but designing good heuristics requires thought.