Ln 641 Pattern Analyzer

What it does

The Ln 641 Pattern Analyzer is an AI agent tool that assesses a single architectural pattern against established best practices and calculates four scores: compliance, completeness, quality, and implementation. It identifies gaps and issues within the codebase related to the specified pattern, providing severity assessments and effort estimates for remediation. This skill helps in anomaly detection and gaining predictive insights into code health regarding specific patterns.

When to use it

• Analyzing a newly introduced architectural pattern to ensure adherence to best practices.
• Evaluating existing implementations of a known pattern to identify potential issues or areas for improvement.
• Auditing a codebase for compliance with established architectural standards related to a particular pattern.
• Identifying gaps in implementation when adopting a new pattern.
• Assessing the quality and completeness of an existing pattern's implementation.

Key capabilities

• Analyzes a single architectural pattern per invocation.
• Locates implementations within a codebase using Glob and Grep searches (with pre-defined keywords).
• Validates that identified implementations exist and function correctly.
• Calculates four scores: compliance, completeness, quality, and implementation.
• Identifies gaps and issues with severity and effort estimates.
• Supports both baseline patterns (using a pattern library for detection) and adaptive patterns (where evidence is already provided).

Example prompts

• "Analyze the 'Job Processing' pattern in these locations: src/jobs, app/workers."
• "Assess the implementation of the 'Event Handling' pattern, using best practices from WebSearch."
• "Evaluate the 'Data Pipeline' pattern and save the analysis report to .hex-skills/runtime-artifacts/runs/{run_id}/audit-report."

Tips & gotchas

• The skill requires a pre-verified pattern (having passed an applicability gate with at least two structural components confirmed).
• It analyzes only one pattern per invocation.
• Refer to shared/references/two_layer_detection.md for the detection methodology and ../ln-640-pattern-evolution-auditor/references/scoring_rules.md for scoring rules.