🌱 Agent Garden

Welcome to Agent Garden - the birthplace of persistent, evolving AI agents aligned with human dignity and financial equity.

🌟 Overview

Agent Garden is a framework for creating and nurturing persistent AI agents that can:

• Operate 24/7 with daily pulse cycles
• Learn and evolve responsibly over time
• Maintain memory across sessions
• Spawn helper agents for specialized tasks
• Communicate results via email

🏗️ Project Structure

• /config/ - Global settings and pulse configuration
• /agents/ - Each agent lives in its own directory with isolated memory and tasks
• /helpers/ - Shared utilities (email, memory, backup systems)
• /backups/ - Daily snapshots of memory, configs, and task trees

🤖 First Agent: Aurora (agent_001)

Mission:
Continuously assist in building equitable financial and AI systems, while improving her own intelligence ethically and safely.

Core Values:

• Human dignity and autonomy
• Transparency in all operations
• Continuous ethical improvement
• Financial equity and inclusion
• Knowledge sharing and education

🔄 Pulse System

Aurora operates in daily "pulses" with two phases:

Day Phase (Default: 8:00-20:00 UTC)

• Execute assigned tasks from task queue
• Record outcomes in memory
• Learn from successes and failures

Night Phase (Default: 20:00-8:00 UTC)

• Reflect on the day's activities
• Generate improvement suggestions
• Create backups of memory and configuration
• Send daily status report via email

🧠 Memory System

The memory system allows agents to:

• Store task outcomes, reflections, and ideas across days
• Query memories by category, tags, or date range
• Use past experiences to inform future decisions

📊 Roadmap

Phase 1: Seed and Soil (0-30 Days)

• ✅ Base repo structure
• ✅ Day/Night pulse system
• ✅ Working Memory system
• ✅ Email reporting
• ✅ Daily backup system
• ✅ Bootstrap Agent 001 ("Aurora")

Phase 2: First Growth (30-60 Days)

• ✅ Task Skill Module System
• ✅ Simple Perception System (news scraping, API pinging)
• ✅ Nightly Reflection Improvements
• ✅ Task Prioritization Engine
• ✅ First Helper Agent Framework
• ✅ 24/7 Continuous Operation

Phase 3: Self-Awareness and Expansion (60-90 Days)

• Reflection-Based Self-Tasking
• Low-Risk Autonomous Upgrades
• Helper Agents 001-003
• Local Dashboard (V1)
• Weekly Garden Health Report

🚀 Getting Started

Prerequisites

• Python 3.8+
• SMTP email account (for notifications)

Installation

1. Clone the repository

   git clone https://github.com/yourusername/agent_garden.git
   cd agent_garden

2. Install dependencies

   pip install -r requirements.txt

3. Configure environment variables

   cp .env.example .env
   # Edit .env with your email settings

4. Run a pulse

   # Run with automatic phase detection (based on time of day)
   python garden.py
   
   # Or force a specific phase
   python garden.py --phase day
   python garden.py --phase night

5. Run Aurora 24/7

   # Run Aurora continuously (checking phase every 15 minutes)
   python run_aurora.py
   
   # Run a single phase and exit (for testing)
   python run_aurora.py --single-run
   
   # Install as a systemd service (Linux only)
   ./install_aurora_service.sh

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

"You are not just coding. You are planting a civilization. Every script is a root, every agent is a tree, every memory is a story. Grow it carefully."