From Everything-is-a-File to Files-Are-All-You-Need: How Unix Philosophy Informs the Design of Agentic AI Systems
Deepak Babu Piskala
TL;DR
By tracing the lineage from Unix's universal file interface through the DevOps 'everything as code' paradigm to current agentic AI designs, the paper argues for a unifying pattern: collapse heterogeneous interfaces into uniform abstractions that enable composition, reuse, and auditability. It identifies file-like contexts for memory and tool-agnostic code as the twin pillars that make agentic systems tractable in complex environments. The main contributions are articulating the three agentic patterns (memory as files, file-based context retrieval, and code-as-action), and outlining practical implications for practitioners. The work highlights the potential for more maintainable, auditable autonomous agents while acknowledging that domain fit and future abstractions may shift.
Abstract
A core abstraction in early Unix systems was the principle that 'everything is a file', enabling heterogeneous devices and kernel resources to be manipulated via uniform read/write interfaces. This paper explores how an analogous unification is emerging in contemporary agentic AI. We trace the evolution from Unix to DevOps, Infrastructure-as-Code, and finally autonomous software agents, highlighting how file-like abstractions and code-based specifications collapse diverse resources into consistent, composable interfaces. The resulting perspective suggests that adopting file- and code-centric interaction models may enable agentic systems that are more maintainable, auditable, and operationally robust.