Add self-evolving meta-layer system (clean extraction from updated main)

[jwm-axoni]

🎯 Overview

This PR introduces a self-evolving meta-layer system that enables Agent Zero to learn and improve from its interactions through automated prompt evolution, tool suggestions, and comprehensive versioning.

✨ Features Added

1. Prompt Evolution Tool (python/tools/prompt_evolution.py)

• Automatically analyzes agent performance and suggests prompt improvements
• Learns from successful and failed interactions
• Generates optimized prompts based on usage patterns

2. Prompt Versioning System (python/helpers/prompt_versioning.py)

• Version control for all system prompts
• Rollback capability to previous prompt versions
• Change tracking and diff visualization
• Backup and restore functionality

3. Tool Suggestion Engine (python/helpers/tool_suggestions.py)

• Analyzes conversation context to suggest relevant tools
• Learns from tool usage patterns
• Improves tool discovery and utilization

4. Meta-Learning API (python/api/meta_learning.py)

• RESTful API endpoints for meta-learning operations
• Prompt analysis and evolution endpoints
• Version management endpoints
• Integration with existing Agent Zero API

5. Auto-Evolution Extension (python/extensions/monologue_end/_85_prompt_evolution.py)

• Automatic prompt evolution after conversations
• Non-intrusive integration via extension system
• Configurable evolution triggers

📚 Documentation

Comprehensive documentation added in docs/meta_learning/:

• QUICKSTART.md - Getting started guide
• README.md - System overview and architecture
• INDEX.md - Documentation index
• DELIVERABLES.md - Feature deliverables
• TEST_ARCHITECTURE.md - Testing strategy
• TEST_SUMMARY.md - Test coverage summary

🧪 Testing

Full test suite included:

• tests/meta_learning/test_prompt_versioning.py - Versioning tests
• tests/meta_learning/manual_test_*.py - Manual testing scripts
• tests/test_meta_learning_api.py - API endpoint tests (478 lines)
• tests/meta_learning/verify_test_structure.py - Test validation

🏗️ Architecture

The meta-layer system follows Agent Zero's design principles:

• ✅ Non-invasive - Uses extension system, no core modifications
• ✅ Modular - Each component is independent and reusable
• ✅ Extensible - Easy to add new meta-learning capabilities
• ✅ Prompt-driven - All behavior configurable via prompts

📊 Files Changed

Added (18 files, 6,431 lines):

• 5 core Python modules
• 7 documentation files
• 4 test files
• 2 prompt templates

Modified: None (zero breaking changes)

🔄 Integration

The system integrates seamlessly with existing Agent Zero features:

• Works with all existing tools and extensions
• Compatible with the current prompt system
• No changes required to existing workflows
• Backward compatible

🚀 Usage Example

# Automatic evolution (via extension)
# Just use Agent Zero normally - prompts evolve automatically

# Manual evolution (via tool)
agent.communicate("Analyze my recent performance and suggest prompt improvements")

# API usage
POST /api/meta_learning/analyze
POST /api/meta_learning/evolve_prompt
GET /api/meta_learning/versions
POST /api/meta_learning/rollback

🎓 Benefits
Continuous Improvement - Agent learns and improves over time
Reduced Manual Tuning - Automated prompt optimization
Better Tool Discovery - Context-aware tool suggestions
Version Safety - Easy rollback if changes don't work
Transparency - Full visibility into prompt evolution

✅ Checklist
Code follows Agent Zero architecture patterns
Comprehensive documentation included
Full test suite provided
No breaking changes
Extension-based integration (non-invasive)
Backward compatible

📝 Notes
This is a complete, production-ready feature that has been tested and documented. It adds powerful meta-learning capabilities while maintaining Agent Zero's core philosophy of extensibility and prompt-driven behavior.

Ready for review! 🎉