Operating systems research and open-source exploration for the AI era. Starting with Linux, we explore how agents and computers can work together.
Explore the research ↗OPEN QUESTIONS. REPRODUCIBLE EXPLORATIONS.
Rethinking system boundaries for the AI era.
How should a system manage the lifecycle, state and resources of agents running ongoing tasks?
How do interrupted tasks recover? How should agents share CPU, memory and tools? How can execution be observable and reproducible?
Ground autonomous execution in permissions that can be understood, revoked and audited.
What are the minimum permissions for a task? How do we isolate untrusted code, record tool calls and let people revoke access at any time?
Build on existing open-source systems to explore kernels, isolation, scheduling and interaction.
What belongs in the kernel and what belongs in user space? How can containers and virtual machines support agents? How do we build reproducible Linux experiments?
These research topics are evolving. Expand a direction to explore its starting questions.
OSFUN is an early-stage research interest site. We explore systems questions for the AI era while building on decades of operating systems work. We start with testable questions and document experimental conditions, processes and limitations. These are directions for exploration, not claims of a finished system or published findings.
Define the scenario and hypothesis. Study relevant system designs and open-source implementations.
Record the environment, steps and observations. Distinguish facts, assumptions and conclusions.
Share notes and reproducible materials, then iterate based on feedback.
Start with open-source foundations: Linux kernel documentation ↗ · QEMU documentation ↗
Join the conversation on system design, Linux experiments, agent runtimes and open-source practice. Bring a question, a research note or a reproducible experiment.