auto-evolution

What problem does it solve?

Auto-Evolution addresses the challenge of agents forgetting learned patterns and re-solving the same issues by continuously capturing interactions, errors, and skill usage to build a persistent knowledge base.

Core Features & Use Cases

• Memory-driven three-layer memory (episodic, semantic, procedural) enables long-term learning from real-world interactions.
• Pattern detection and draft generation convert repetitive experiences into reusable skills.
• Use cases include prompting automatic draft creation for recurring errors or workflows, reducing manual maintenance and accelerating improvement.

Quick Start

Enable the evolution hooks in your environment and start using skills as usual; the system will silently capture usage, analyze patterns at session end, and propose drafts for new skills in memory/drafts.