continuous-learning

You can automatically extract reusable patterns from Claude Code sessions by enabling a Stop hook that evaluates the session when it ends and saves detected patterns to the local learned-skills directory.

Yes, you can configure the minimum session length and extraction threshold options to tailor learning, ensuring the Stop hook only saves patterns from sessions that meet your defined criteria.

Extracted learning patterns are saved as learned skills under the ~/.claude/skills/learned/ directory on your local machine after the Stop hook runs at the end of a session.

A Stop hook captures learned skills by running at the end of a Claude Code session, identifying reusable patterns from the interaction, and writing them to the local learned-skills directory for future reuse.

The best way to turn transient interactions into persistent skills is using an automated Stop hook to detect reusable patterns at session end and save them to ~/.claude/skills/learned/ for future reuse.

Pattern detection is constrained by configurable extraction thresholds and minimum session lengths, meaning sessions that do not meet these criteria will not have learned skills extracted or saved.