pattern-learning

Pattern learning detects successful task execution patterns by classifying tasks, extracting context like languages and frameworks, then stores them in a `.claude-patterns/` directory. This creates a project-specific knowledge base that improves future skill selection and decision-making.

Yes. Pattern learning applies to refactoring, bug fixes, feature development, optimization, documentation, testing, and security tasks across diverse languages and frameworks, automatically extracting context to identify and categorize successful approaches.

The skill calculates a quality score for each task outcome and stores patterns with skill effectiveness metrics. These historical success rates drive the skill auto-selection algorithm, recommending optimal skills for similar future tasks and accelerating development cycles.

The skill automatically stores the approach, skills used, quality score, and task history as a new pattern in the centralized repository, making that solution available for recommendation on similar future tasks.

No. Pattern learning builds your knowledge base from scratch, starting with your first successful task. As more patterns accumulate, recommendations and skill selection become increasingly data-driven and accurate.

Pattern learning tracks skill effectiveness metrics, quality scores, task categorization, extracted context (languages and frameworks), and execution history. These metrics feed the auto-selection algorithm to recommend optimal skills for future tasks.