🚀 OpenAgent

A sophisticated multi-agent LLM orchestration system featuring meta-cognitive capabilities, adaptive memory management, and deep learning-based evolution.

OpenAgent is not just a chatbot; it is an intelligent framework built on the principle that true AI assistance comes from genuine attention and shared growth. It learns from you, notices your patterns, and builds a living model of your needs over time.

---

Guide

• Meta-Cognitive Intelligence
• Architecture
• Adaptive Memory System
• Key Features
• Installation & Setup
• Repository Structure
• Contributing
• License

🧠 The "Secret Sauce": Meta-Cognitive Intelligence

Unlike traditional chatbots that provide generic, forgettable interactions, OpenAgent uses a Meta Agent to analyze every message before the main response is even generated.

• Intent & Tone Detection: Understands what you really want and notices emotional shifts (e.g., frustration or stress).
• "Aha Moments": Identifies insights to share, proving the system is paying genuine attention.
• Learning Opportunities: Actively asks questions to learn your specific workflows, like your preference for TDD or specific design patterns.
• Mental Health Tracking: Observes stress patterns over time to provide empathetic context.

---

🏗️ Architecture

This system uses a unified adaptive pipeline with three execution paths:

• fast for low-complexity turns
• standard for normal multi-step turns
• deep for complex turns and sub-agent orchestration

---

📚 Adaptive Memory System

The system manages three core Markdown-based memory files that evolve through interaction:

• identity.md: The assistant’s core values and principles.
• soul.md: Philosophical stance and deeper purpose.
• user.md: A growing profile of you, including your expertise, preferences, and mental health trajectory.

Smart Loading: To save tokens and improve performance, memory files are only loaded when the Meta Agent determines they are relevant to the context.

---

🔧 Key Features

• Universal LLM Support: Works with OpenAI, Anthropic (Claude), Groq, Mistral, Together AI, OpenRouter, DeepSeek, and custom endpoints.
• Powerful Toolset: Built-in tools for shell_command execution (with approval), read_file, edit_file, regex_search, and curl_request.
• Resilient Execution: Includes automatic 429 rate-limit handling (30s retry), loop guards for failing tools, and 0.5s delays between API calls.
• Persistent Sessions: SQLite-backed history allows you to resume, rename, or delete chats.

---

🛠️ Installation & Setup

1. Clone and Install

# Open the folder
cd OpenAgent

# Install dependencies
pip install -r requirements.txt

2. Launch the System

python main.py

Open http://localhost:8000 in your browser to begin.

3. Recommended Configuration

Go to "Settings" to configure your providers. Best practice suggests:

• Meta Agent: Fast & Cheap (e.g., Groq Llama 3, Mixtral, or GPT-3.5).
• Main Agent: Smart & Capable (e.g., Claude 3.5 Sonnet, GPT-4, or Mistral Large).
• Sub-Agents: Fast & Focused (e.g., GPT-3.5 or Llama 3).

---

📂 Repository Structure

• main.py: FastAPI backend.
• index.html: Frontend interface.
• prompts/: Core instructions for Meta and Main agents.
• memory/: Persistent memory files (identity.md, soul.md, user.md).
• sessions.db: SQLite database for chat history.

---

🤝 Contributing

OpenAgent is a framework for intelligent, adaptive assistance. We welcome contributions:

• Adding new tools or providers.
• Enhancing meta-analysis logic.
• Modifying agent prompts for different interaction styles.

---

📜 License

This project is licensed under the MIT License.

---

"The best assistant isn't the one that knows everything - it's the one that really knows you."