pattern-learning_skill

Overview

This skill enables autonomous pattern recognition, storage, and retrieval at the project level with self-learning capabilities for continuous improvement. It builds a local knowledge base of successful task approaches, recommends skills and agents based on historical outcomes, and continuously adjusts recommendations through self-assessment. The design is privacy-first with 100% local processing and production-ready safety checks.

How this skill works

During task execution the system extracts task type, context (languages, frameworks, project structure), and complexity, then monitors decisions, skills loaded, and execution metrics. After completion it runs a quality assessment, computes a quality score, extracts lessons learned, and safely stores the new pattern in a local pattern database (.claude-patterns). A selection algorithm queries stored patterns to recommend an ordered list of skills and agents for similar future tasks.

When to use it

• Automating skill selection for recurring development tasks (refactoring, bug-fix, feature work).
• Improving code-review and PR automation by reusing proven approaches and agents.
• Bootstrapping test and security strategies based on past successful executions.
• Tracking long-term trends in skill and agent effectiveness for process improvement.
• Maintaining an auditable, local history of task outcomes and lessons learned.

Best practices

• Validate inputs and project context before creating or updating patterns to avoid noise.
• Run full quality assessments (tests, linting, docs) after tasks to produce meaningful scores.
• Prefer safe defaults when the pattern database is missing or content is invalid.
• Use similarity thresholds to avoid merging distinct approaches; increment reuse only for truly similar patterns.
• Periodically review and deprecate low-performing skills or patterns based on trend analysis.

Example use cases

• Refactor an authentication module: auto-select code-analysis and quality-standards based on prior high-quality refactors.
• Add unit tests for payment processing: recommend testing-strategies and quality-standards tailored to language and domain.
• Continuous pipeline: store each task outcome to track quality score trends and adjust future skill weights.
• Security hardening: detect security-critical contexts and favor additional validation steps from previous successful runs.