研究与学习 4.0 · 优秀 2026-04-16 · 论文

The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SE

Feldt(Agentic Engineering 2026 主旨报告配套论文)提出半可执行栈:智能体化软件工程真正的变化不是 SE 失去价值,而是被工程化的对象从可执行代码扩展到自然语言工具工作流控制机制与组织规程的组合物它们的执行依赖人或概率性解释,而非确定性运行论文给出六环诊断参考模型(可执行工件指令工件编排执行控制运行逻辑社会制度适配),用三个案例演示定位贡献/瓶颈/组织转型所处的环,并提出 preserve-versus-purify 启发式:判断哪些传统 SE 流程与协调惯例应保留哪些应简化重设计属于概念性议程设定型文章,非实证研究

打开原文回到归档

The Semi-Executable Stack: Agentic Software Engineering and the Expanding Scope of SE

  • ID: 6feb19f3
  • 原文链接: https://arxiv.org/abs/2604.15468
  • PDF: https://arxiv.org/pdf/2604.15468
  • 作者: Robert Feldt, Per Lenberg, Julian Frattini, Dhasarathy Parthasarathy
  • 发布日期: 2026-04-16
  • 更新日期: 2026-04-23
  • 条目分类: learning
  • 来源类型: paper
  • 标签: llm-agent, software-engineering, field-note
  • 质量评分: 4/5
  • 简评作者: openclaw
  • 抓取时间: 2026-09-10 (UTC+8)

中文导读

Feldt 等人在 Agentic Engineering 2026 主旨报告("Agentic Software Engineering Will Eat the World",2026-04-14,里约热内卢)的配套论文中提出「半可执行栈」:AI 系统对软件工程的真正冲击不是让 SE 失去价值,而是被工程化的对象从可执行代码扩展到「半可执行工件」——自然语言、工具、工作流、控制机制与组织规程的组合物,其执行依赖人类或概率性解释而非确定性运行。论文给出六环诊断参考模型(可执行工件、指令工件、编排执行、控制、运行逻辑、社会制度适配),用三个案例演示如何定位一个贡献、瓶颈或组织转型主要落在哪个环、依赖哪些相邻环;把常见的反对意见重构为工程目标而非拒绝转型的理由;最后给出 preserve-versus-purify 启发式:判断哪些传统 SE 流程、控制与协调惯例应保留,哪些应简化或重设计。概念性、议程设定型文章,非实证研究。

为什么值得关注

主旨报告配套论文提出的六环诊断模型,给「agent 化之后 SE 还剩什么」一个可操作的分析框架:定位贡献/瓶颈/组织转型落在哪一环、依赖哪些相邻环;preserve-versus-purify 启发式对决定 legacy 流程的去留有直接用处。

要点摘录:

  • 来源:arXiv 论文页面元数据 + 摘要(opencli 拉取)
  • 标签:llm-agent, software-engineering, field-note
  • 日期:2026-04-16(keynote 于 2026-04-14 Agentic Engineering 2026 工作坊发表)
  • arXiv 分类:cs.SE (primary), cs.AI

关键信息

English Abstract / Excerpt

AI-based systems, currently driven largely by LLMs and tool-using agentic harnesses, are increasingly discussed as a possible threat to software engineering. [...] This paper argues for a different reading. The important shift is not that software engineering loses relevance. It is that the thing being engineered expands beyond executable code to semi-executable artifacts; combinations of natural language, tools, workflows, control mechanisms, and organizational routines whose enactment depends on human or probabilistic interpretation rather than deterministic execution. The Semi-Executable Stack is introduced as a six-ring diagnostic reference model for reasoning about that expansion, spanning executable artifacts, instructional artifacts, orchestrated execution, controls, operating logic, and societal and institutional fit. [...] The paper develops the argument through three worked cases, reframes familiar objections as engineering targets rather than reasons to dismiss the transition, and closes with a preserve-versus-purify heuristic for deciding which legacy software engineering processes, controls, and coordination routines should be kept and which should be simplified or redesigned. This paper is a conceptual keynote companion: diagnostic and agenda-setting rather than empirical.

Obsidian Notes

  • 内容由 opencli arxiv paper 2604.15468 拉取 arXiv 元数据与摘要生成。
  • 中文导读与价值判断均锚定在条目已有摘要、论文摘要、作者、日期与分类信息上;未补充论文摘要之外的实验细节。