A universal self-improving agent that learns from ALL skill experiences. Uses multi-memory architecture (semantic + episodic + working) to continuously evolv...
After any skill completes, this agent automatically:
1. Analyzes what happened
2. Extracts patterns and insights
3. Updates relevant skill files
4. Logs to memory for future reference
5. Reports summary to user
๐ Tips & Best Practices
DO
โ Learn from EVERY skill interaction
โ Extract patterns at the right abstraction level
โ Update multiple related skills
โ Track confidence and apply counts
โ Ask for user feedback on improvements
โ Use evolution/correction markers for traceability
โ Validate guidance before applying broadly
DON'T
โ Over-generalize from single experiences
โ Update skills without confidence tracking
โ Ignore negative feedback
โ Make changes that break existing functionality
