Neo.mjs

•
3.3k
•

Empowers a self-evolving AI engineering team to inhabit and autonomously maintain live applications and codebases.

Neo.mjs is a self-evolving software organism that acts as an end-to-end AI engineering team, designed to inhabit and autonomously maintain live applications and codebases. It leverages a swarm of AI models (Claude, Gemini, GPT) that collaboratively read and review each other's reasoning via shared memory and Active Hybrid GraphRAG. Through its Neural Link interface, this AI swarm can inspect and mutate live application states, effectively turning conversational UIs into agents that collaborate within the application itself. It manages the full engineering lifecycle, from ideation and building to cross-reviewing, self-healing, and continuous improvement through mechanisms like DreamService.

Key Features

01Neural Link for live application inhabitation and real-time state mutation.
02Autonomous engineering lifecycle including ideation, building, and self-healing.
03Multi-threaded application engine for high-performance UI and data processing.
04Self-evolving AI engineering team with a cross-model swarm intelligence.
05Persistent Memory Core and Native Edge Graph for queryable reasoning.
063,190 GitHub stars

Use Cases

01Developing multi-agent AI systems requiring persistent memory and runtime-mutable application substrates.
02Building high-performance, multi-window enterprise applications (e.g., financial trading platforms, IDEs).
03Autonomously maintaining and evolving complex codebases with an AI engineering team.