pattern-detection_skill

Overview

This skill detects patterns, anomalies, and trends in code and data to help surface code smells, security weaknesses, and recurring issues. It combines regex matching, AST analysis, and statistical anomaly detection to provide actionable findings. Reports prioritize severity and recommend verification steps to reduce false positives.

How this skill works

The skill scans source files and text using configurable regex rules and AST inspections to find insecure constructs, hardcoded secrets, long or duplicated functions, and poor error handling. For numeric or time-series data it applies z-score, IQR, and moving-average trend analysis to detect outliers and directional changes. Results are aggregated into a severity-ranked report that highlights files/lines, pattern types, and suggested next steps.

When to use it

• During code review to pre-filter common code smells
• Before or during security audits to find vulnerability patterns and hardcoded secrets
• When preparing a refactor to locate long or duplicated functions
• Monitoring logs or datasets for sudden anomalies or spikes
• Analyzing product metrics for trend direction and volatility

Best practices

• Start with lightweight, conservative regexes to limit false positives
• Verify matches in context before making code changes
• Prioritize findings by severity: HIGH (security), MEDIUM (maintainability), LOW (style)
• Combine static patterns with AST checks to reduce syntactic false matches
• Document and re-run scans after fixes to confirm resolution

Example use cases

• Scan a repository to list hardcoded API keys, eval/exec usage, and empty catch blocks
• Identify functions exceeding a configurable line threshold for targeted refactoring
• Detect sensitive data (emails, credit cards, SSNs) leaked in logs or datasets using regex rules
• Run z-score or IQR on metric series to flag anomalous timepoints for investigation
• Compare 7-day and 30-day moving averages to report trend direction and average growth