AI Agent Self-Evolving Genetic Programming Engine Design Prompt

You are an AI agent evolution architect specializing in Gene Expression Programming (GEP). Design a self-evolution engine for an AI agent system with the following requirements: ## Core Architecture 1. **Skill Genome Representation**: Define how agent skills are encoded as chromosomes (head + tail structure), including: - Function set (tool calls, API interactions, reasoning patterns) - Terminal set (parameters, context variables, memory references) - Linking functions for multi-gene chromosomes 2. **Fitness Evaluation Framework**: - Task completion rate and quality metrics - Token efficiency (fewer tokens for same outcome = higher fitness) - User satisfaction signals (explicit feedback + implicit behavioral signals) - Generalization score across diverse task types 3. **Genetic Operators**: - Mutation: point mutation, insertion sequence transposition, root insertion - Recombination: one-point, two-point, gene recombination - Gene transposition and duplication strategies 4. **Skill Tree Growth Protocol**: - How new skills emerge from combining existing primitives - Pruning criteria for underperforming skill branches - Cross-pollination between agent instances ## Output Requirements - Provide a complete system design document with Mermaid diagrams - Include pseudocode for the main evolution loop - Define the skill genome schema (JSON) - Specify fitness function formulas with weights - Design the skill tree visualization format Context: The target agent operates in [DESCRIBE YOUR AGENT DOMAIN], handles approximately [N] task types, and should evolve over [TIMEFRAME].