Re: AGENTS.md and CLAUDE.md Are Concessions to Proprietary Services, Not Defenses Against Them
Yuri Khan <[email protected]> Tue, 4 Aug 2026 14:09:41 +0700
| Newsgroups | gmane.emacs.devel |
|---|---|
| Message-ID | <CAP_d_8X_zOPUnv7E8pRLW1NXeAQHWdT97qrgSpsYZzU42w+WLA@mail.gmail.com> |
On Tue, Aug 4, 2026 at 1:36=E2=80=AFPM Jean Louis <[email protected]> wrote: > This is not FUD. It's a principled question: why should GNU adopt a > convention created by a proprietary company That=E2=80=99s what GNU is, historically: A reimplementation of conventions created by a proprietary company. (UNIX) > Let me take example of OpenCode; > > Ignore AGENTS.md and and improve my-package.el You=E2=80=99re talking like you assume the LLM to actively look for and rea= d that file. That is not what happens. The harness (Pi, OpenCode, whatever) reads the file and appends its contents to the system prompt that is given to the LLM as its context starter. The output of the LLM then may or may not be affected by the presence of that prompt fragment. Yes, there are models that are better or worse at following prompts. From the point of view[1] of the LLM, you[2] gave it an instruction to not modify any files, then another instruction to ignore an unrelated file, and then a third instruction to improve a file. The first and third conflict. The second is a slight hint to favor the third. [1]: expression used as analogy only; LLMs do not actually have points of v= iew. [2]: the collective entity consisting of the harness and the user. > And finally, why would Emacs User necessarily use OpenCode in the shell? Who said anything about shell? Many harnesses implement a way to use them as a middle tier between the LLM and the UI. There exist UIs implemented as Emacs modes. > Why assume that LLM generated code will be generated over those software > pieces which use AGENTS.md? Because that=E2=80=99s the reasonable thing to assume: Over time, most harnesses will converge on a shared convention for project-local customizations, and AGENTS.md seems to be the most likely. > Majority of Emacs packages for LLM code generation do not use AGENTS.md, File bugs at them. > And you cannot prove that AGENTS.md will work. Of course you can=E2=80=99t. You can=E2=80=99t prove the same for a human r= eading CONTRIBUTE, either. Basically as a maintainer of a GPL-licensed project you can=E2=80=99t enfor= ce any policy on the individual user, until and unless they start redistributing it.