Functional Graph AGI Prototype: Modular Reasoning Engine with Curiosity, Abstraction, and User-Guided Learning

This project presents a working AGI prototype designed as a functional graph of cognitive modules. Each module is memory-aware, explainable, and equipped with trust tracking. The system supports symbolic reasoning, semantic conflict detection, user-guided teaching, hypothesis generation via a local LLM (Ollama), and module abstraction.

The prototype is implemented in Python, includes a Streamlit-based interface for interactive exploration, and supports dynamic expansion via user interaction and automated training. It is suitable for research in explainable AI, reasoning architectures, and modular AGI systems.

Live Demo:
A non-interactive video demonstration and user interface overview are available at:
http://limitgraph.com/#rec1088140126

In addition, a public API is available for testing the system’s reasoning capabilities in real time.
Instructions and an interactive interface are provided here:
http://api.limitgraph.com:8000/docs

Note:
The full source code of this AGI prototype is currently private due to its experimental and potentially commercial nature.
If you are interested in accessing the repository for academic research, peer review, or collaboration, please contact the author:

Email: alex21259alex@gmail.com
Email (for grants/investors): info@limitgraph.com
Telegram: @Alex_larinov

Access may be granted on a case-by-case basis under the terms of the included Non-Commercial License.