OSSARTH
An LLM-native operating system research prototype.
Overview
OSSARTH explores how language models could translate user intent into task plans and coordinate tools over a Linux-based system. The project includes an architectural proposal, orchestration components, and a simulated task-management prototype.
My contribution focused on hierarchical agent orchestration, MCP tool integration, and the task-management interface. I also co-authored the research manuscript.
Proposed architecture
The architectural proposal uses language models to interpret intent and plan tasks, with hierarchical agents coordinating execution through tools. MCP provides an interface for tool integration. Scheduling and resource-management workflows are part of the proposed design over the Linux kernel.
The architectural scope describes a research direction; the implemented components demonstrate a narrower interaction and orchestration workflow.
Implemented scope
The prototype includes orchestration components and a simulated task-management interface with a seven-panel system monitor. This provides a way to explore task coordination and inspect simulated system activity.
This is a research prototype. It is not a replacement Linux kernel, a production operating system, or evidence of a new scheduler outperforming existing systems.
Evidence and current status
The manuscript records the project’s proposed design and prototype work. It was submitted for publication according to my résumé. No acceptance or peer-reviewed publication is claimed here.
No reproducible operating-system performance measurements are presented on this page. Broader architectural capabilities remain distinct from the implemented simulation.