continuous-learning

Pattern extraction from Claude Code sessions identifies recurring strategies and best practices in your transcript, then saves them as learned skills to ~/.claude/skills/learned/ for future reuse after each session ends.

Session analysis surfaces meaningful patterns by evaluating transcript content to detect recurring strategies, decision-making approaches, and best practices that recur across workflows, converting them into reusable skill artifacts.

A minimal configuration file controls thresholds and paths for pattern storage. Learned patterns persist to ~/.claude/skills/learned/ by default, with configurable directories and evaluation thresholds to refine what qualifies as reusable.

Yes. The skill requires a Stop hook mechanism and session transcript access, enabling continuous pattern evaluation across sessions so learned patterns accumulate and remain available for reuse in future workflows.

You need Claude Code session transcript access, a Stop hook mechanism to trigger pattern evaluation at session end, and a configuration file specifying thresholds and storage paths—no external dependencies required.