auto-evolve

What problem does it solve?

Automates the growth and governance of evolving shadow-organisms (skills/hooks/agents) by turning user feedback memories and dream signals into adaptive entities governed by a multi-dimensional Fitness Oracle and a signed promotion FSM; it ensures stable organisms are auto-registered into the Claude Code skill loader and tracked across invocations with attribution counters.

Core Features & Use Cases

• Auto-evolve kernel converts feedback memories and dream signals into shadow organisms that adapt over time.
• Signed promotion FSM governs transitions between statuses (shadow, proposal, stable, archived) with auditable promotion, extension, and decay rules.
• DreamEvidence feed provides the Oracle with read-only observations to influence promotions; per-organism session links and FitnessScores feed status panels and analytics.
• Auto-registration into the Claude Code skill loader and seamless integration with /evolve-status workflows for monitoring and management.

Quick Start

Invoke the auto-evolve engine to begin evolving shadow organisms from your feedback memories and dream signals.