continuous-learning

To automatically extract reusable patterns from Claude Code sessions, you can use a continuous learning skill that evaluates completed sessions via a Stop hook, detects meaningful patterns, and saves them as learned skills for future reuse.

Continuous learning in Claude Code is an automated pattern extraction process that analyzes completed sessions, detects reusable patterns based on configurable types and thresholds, and stores them in ~/.claude/skills/learned/ for future reuse.

Yes, you can customize pattern types and thresholds for session analysis. The skill supports configurable pattern types and thresholds, allowing you to control what meaningful patterns are extracted and saved from your Claude Code sessions.

To save learned skills from Claude Code logs, the skill automatically analyzes your session data and stores the extracted patterns in the ~/.claude/skills/learned/ directory, making them available for quick reuse in future sessions.

Yes, a Stop hook is required to enable automatic pattern extraction. The skill runs via a Stop hook after each Claude Code session to automatically evaluate completed sessions and extract reusable patterns without manual intervention.

The best way to avoid re-deriving patterns from each Claude Code session is to use an automated continuous learning skill that extracts reusable patterns at session end and saves them as learned skills, eliminating repetitive manual pattern identification.