continuous-learning

To save reusable patterns automatically, this tool evaluates completed Claude Code sessions and extracts recurring patterns into a local skills library. It requires a Stop hook integration to trigger the evaluation at the end of each session.

Continuous learning in Claude Code automatically evaluates completed sessions to identify recurring patterns and saves them locally. It works by applying a Stop hook to trigger session evaluation, extracting best-practice patterns into a growing repository for future workflows.

Yes, extracting learned skills requires a configured minimum session length to ensure meaningful patterns are captured. You also need a Stop hook integration to trigger evaluation and a local directory at ~/.claude/skills/learned/ to store the extracted patterns.

When automating Claude Code session evaluation, extracted patterns are saved directly to a local learned-skills directory at ~/.claude/skills/learned/. This builds a growing repository of reusable patterns accessible for future debugging tasks and learning workflows.

Automated pattern extraction requires a Stop hook integration and a minimum session length. It is not suitable for very short sessions that fail to meet the minimum length threshold, as these sessions will not trigger the evaluation needed to save learned skills.