The maintainers of the Linux kernel are evaluating a formal method to manage the rising volume of automated code contributions. A patch set submitted on September 29, 2026, proposes introducing a dedicated AGENTS.md file—a clear, machine-readable specification designed to guide AI and Large Language Model (LLM) agents that submit code to the project.
As reported by Phoronix, the kernel has long received patches generated by AI tools, but has lacked a standardized document to communicate expectations directly to these non-human contributors. The proposal aims to address this operational gap by creating a direct, structured guideline for the "AI developer" workflow.
The core issue is one of practical maintenance. While the kernel has extensive documentation for human contributors, such as submitting-patches.rst, these documents are often written in prose that automated systems struggle to parse. This results in a high volume of low-quality or non-compliant patches, consuming valuable time from human reviewers who act as the final gatekeepers of quality.
The proposed AGENTS.md would serve as a targeted filter. Its purpose is to outline mandatory standards for automated submissions and explicitly discourage contributions that add minimal value, thereby improving the overall signal-to-noise ratio in the project's patch queue. This isn't a theoretical exercise; it's a direct response to a scaling problem that threatens to slow kernel development.
The discussion marks a significant moment in open-source governance. If adopted, it would represent one of the most prominent formal acknowledgments of AI as a distinct category of participant in a major software project, setting a precedent for how established communities might adapt their processes to coexist with automated tools.
However, several open questions surround the proposal. Its effectiveness depends on AI tool developers choosing to program their agents to read and comply with the document—a factor outside the kernel's direct control. There are also questions about long-term maintenance: who will update the file as AI capabilities and kernel processes evolve?
Ultimately, the Linux kernel's consideration of AGENTS.md reflects a broader trend. As AI coding assistants become commonplace, open-source projects are being forced to rethink contribution workflows and governance structures to accommodate this new form of collaboration. The outcome of this debate may well influence how countless other software communities adapt to an AI-assisted future.
Linux內核維護者正評估一套正式方法，以管理日益增多的自動化代碼貢獻。於2026年9月29日提交的一組補丁提案，建議引入專用的 AGENTS.md 文件——一份清晰、機器可讀的規範，旨在指導向該項目提交代碼的人工智能與大型語言模型（LLM）代理程式。
據Phoronix報導，內核長期以來接收由人工智能工具生成的補丁，但一直缺乏標準化文件來直接向這些非人類貢獻者傳達期望。此提案旨在解決這一操作缺口，為「AI開發者」工作流程創建直接、結構化的指導方針。
核心問題在於實際維護需求。雖然內核已為人類貢獻者提供大量文件，例如 submitting-patches.rst，但這些文件通常以散文形式撰寫，自動化系統難以解析。這導致大量低品質或不合規範的補丁，消耗了擔任最終品質把關者的人類審查者的寶貴時間。
提案中的 AGENTS.md 將充當針對性過濾器。其目的在於概述自動化提交的強制性標準，並明確遏止貢獻附加價值極低的內容，從而提升項目補丁佇列的整體信噪比。這並非理論演習，而是對威脅內核開發進度的擴展性問題的直接回應。
此次討論標誌著開源管治的重要時刻。若獲採用，這將成為大型軟件項目中，人工智能作為獨特參與者類別最顯著的正式認可之一，為現有社群如何調整流程以與自動化工具共存樹立先例。
然而，圍繞此提案仍有多個待解問題。其有效性取決於人工智能工具開發者是否願意編寫讓其代理程式閱讀並遵守該文件的程式碼——這因素超出內核的直接控制範圍。長期維護的問題亦懸而未決：隨著人工智能能力與內核流程的演進，由誰來更新該文件？
終究，Linux內核對 AGENTS.md 的考量反映了更廣泛的趨勢。隨著人工智能編碼助手日益普及，開源項目被迫重新思考貢獻工作流程與管治結構，以適應這種新型協作形式。此番辯論的結果，很可能影響無數其他軟件社群如何適應人工智能輔助的未來。