memory-to-skill-crystallizer

To convert daily memory notes into reusable skills, a crystallizer scans memory/YYYY-MM-DD.md logs to detect recurring patterns and generates new local skills under skills/local. It reduces manual extraction work and prevents important lessons from getting lost across sessions.

Pattern detection for knowledge management scans memory files for repeated issues and extracts common patterns. This mechanism identifies recurring lessons in daily logs that are suitable for conversion into reusable skills.

To generate local skills from recurring issues in memory files, run the crystallizer to scan logs and create entries in skills/local/<pattern>-recovery. Each generated skill includes a SKILL.md frontmatter with a name and description and a basic template.

Yes, skill generation requires memory logs formatted as memory/YYYY-MM-DD.md files. The deterministic pattern detection scans these specific daily markdown logs to extract recurring patterns and create corresponding local skills.

The best way to register and discover generated local skills is through automatic listing under skills/local. The crystallizer handles registration by listing generated skills in this directory for quick discovery and reuse.