local-aios 0.0.2

An AI Operating System for local LLMs — register apps, and let AI agents use them.

local-AIOS

An AI Operating System for local LLMs — register apps, and let AI agents use them.

local-AIOS is a base system that turns a locally-running LLM into the core of an operating-system-like environment. Apps are registered into the system, and AI agents call them as needed — much like programs running on top of an OS kernel. Everything runs on your own machine.

Install name: uv add local-aios · Import name: import local_aios

---

Why local-AIOS?

Most "AI agent OS" projects assume a cloud LLM. local-AIOS is built the other way around: local-first. Your models, your apps, and your data stay on-device.

• Local LLM at the core. Designed for models you run yourself (e.g. via Ollama, llama.cpp, or any OpenAI-compatible local endpoint). No cloud dependency required.
• Apps as first-class citizens. Register apps into the system; agents discover and invoke them through a uniform interface.
• A base system to build on. local-AIOS is the foundation. Higher layers — including a GUI — are designed to sit on top of it.
• On-device by default. Your prompts, files, and app data never have to leave your machine.

---

Status

⚠️ Early development. APIs are not yet stable and may change. The base system is the current focus; the GUI layer is planned for a later stage.

---

Installation

uv add local-aios

How is this different from other "AIOS" projects?

There are several projects exploring the "AI as an operating system" idea. local-AIOS is specifically a local-first base system: it targets self-hosted LLMs running on-device, and is intended as a foundation that a GUI and higher-level apps are built on top of. If you're looking for a cloud-oriented agent platform, other projects may fit better — local-AIOS is for keeping the whole stack on your own machine.

---

Contributing

Contributions are welcome. Since the project is in early development, opening an issue to discuss a change before submitting a large pull request is appreciated.

---

License

Licensed under the Apache License 2.0.