Skip to content
Preprint

Reproducibility in the Age of Agentic AI: Context Engineering at the Timescale of a Codebase

Sep 2026 · 0 citations · 9 references
Computer Science

TL;DR

It is argued that agents lower the cost of maintaining tests, commit histories, repository structure, instructions, and decision records while making their benefits immediate while making their benefits immediate.

Abstract

Reproducible research practices are context engineering for AI coding agents. I argue that agents lower the cost of maintaining tests, commit histories, repository structure, instructions, and decision records while making their benefits immediate. Researchers remain responsible for verifying these artifacts and the scientific judgments they encode.

View source

Similar papers

Case report 2026

Agentic AI for Academic Research

This seminar explores how agentic AI can be integrated into academic research as a disciplined, verifiable, and human-led method rather than a generic productivity tool. It provides practical frameworks for responsible use, workflow verification, and multi-agent research design across literature work, analysis, writing...

M. Zyphur · 0 citations
Sep 2026

The End of the Coder?

As AI agents transition from simple coding assistants to end-to-end engineers, the profession is shifting from writing syntax to auditing complex systems.

Logan Kugler · 0 citations
Case report 2026

Building Autonomous AI Agents for Scientific Research

This workshop explores how autonomous AI agents can support scientific research in R, from conceptual distinctions between chat systems and agentic workflows to practical deployment with open-source tools (OpenCode). It is designed for researchers who want to use AI critically, securely, and effectively in data-intensi...

R. Maestri · 0 citations
Case report 2026

AI as a Research Method

This seminar explores how to treat artificial intelligence as a configurable research method rather than a chatbot or generic software tool. It provides a rigorous framework for validating AI-enabled workflows, designing defensible studies, and making scientific claims that remain consistent with the available evidence...

M. Zyphur · 0 citations
#artificial intelligence Preprint Sep 2026

The Compiler May Read It, the Agent May Not: Keeping Part of a Research Code Away from a Coding Agent

The compiler must read modules a physics-based solver cannot build without; the coding agent must not read that intellectual property. The harness does not ship that rule. We classified fifteen read routes against a container, permission rules and a sandbox. None of the three can tell which program is reading.

Shobhan Roy · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.