Обсуждение: Feature request: Add a llms.txt file to the docs for AIs to learn
Feature request: Add a llms.txt file to the docs for AIs to learn
От
Adrián Cuadrado Chavarría
Дата:
Some projects like [NextJS](https://nextjs.org/docs/llms-full.txt), [Stripe](https://docs.stripe.com/llms.txt), [BetterAuth](https://www.better-auth.com/llms.txt) and [Zod](https://zod.dev/llms.txt) already provide this.
With this file we can make sure that the AI model fetching it is always using the latest features and versions of postgresql.
With this file we can make sure that the AI model fetching it is always using the latest features and versions of postgresql.
On Sun, 2025-11-02 at 00:47 +0100, Adrián Cuadrado Chavarría wrote: > Some projects like [NextJS](https://nextjs.org/docs/llms-full.txt), [Stripe](https://docs.stripe.com/llms.txt), [BetterAuth](https://www.better-auth.com/llms.txt)and [Zod](https://zod.dev/llms.txt) already provide this. > > With this file we can make sure that the AI model fetching it is always using the latest features and versions of postgresql. > > https://llmstxt.org/ Why should we invest effort to feed the machine? I don't think that that improves life for anybody, except perhaps for the owners of said LLMs. Yours, Laurenz Albe
Re: Feature request: Add a llms.txt file to the docs for AIs to learn
От
Adrián Cuadrado Chavarría
Дата:
> Why should we invest effort to feed the machine?
Great news: you don't have to. I don't mind investing the effort myself so no one else has to. It's just a matter of creating a script that would concatenate all markdown files of the documentation and run it automatically whenever a change is made in any document.
> I don't think that that improves life for anybody, except perhaps for the owners of said LLMs.
Users of LLMs would appreciate having a feature like this one. But really, it's just a feature request, not a command or an order. If more people agree that it wouldn't be a good idea to have something like this then I'll just stop pushing it. If some others think this would be a nice thing to have then maybe it would be worth it to make this real.
On Sun, Nov 2, 2025 at 9:58 AM Laurenz Albe <laurenz.albe@cybertec.at> wrote:
On Sun, 2025-11-02 at 00:47 +0100, Adrián Cuadrado Chavarría wrote:
> Some projects like [NextJS](https://nextjs.org/docs/llms-full.txt), [Stripe](https://docs.stripe.com/llms.txt), [BetterAuth](https://www.better-auth.com/llms.txt) and [Zod](https://zod.dev/llms.txt) already provide this.
>
> With this file we can make sure that the AI model fetching it is always using the latest features and versions of postgresql.
>
> https://llmstxt.org/
Why should we invest effort to feed the machine? I don't think that that improves
life for anybody, except perhaps for the owners of said LLMs.
Yours,
Laurenz Albe
The convention here is to add responses inline, after the quoted content being responded to.
> Why should we invest effort to feed the machine?Great news: you don't have to. I don't mind investing the effort myself so no one else has to. It's just a matter of creating a script that would concatenate all markdown files of the documentation and run it automatically whenever a change is made in any document.
The current build for the documentation doesn’t produce Markdown.
> I don't think that that improves life for anybody, except perhaps for the owners of said LLMs.Users of LLMs would appreciate having a feature like this one.
People are going to use LLMs - I have no problem with my volunteer efforts to improve the documentation to be read and referenced by LLMs so that they at least have correct content to work from and save me the effort of having to further correct mistakes or deal with misinformed people. Not that I’m volunteering to do the work. It LLMs seem like a natural pairing with open source principles. Our documentation is already known to be an official source for attribution purposes. Though I’m not totally convinced about it ability to handle our versioning policies correctly if we rely on generalized models scraping us.
David J.
> On 2 Nov 2025, at 00:47, Adrián Cuadrado Chavarría <adriancuadradochavarria97@gmail.com> wrote: > > Some projects like [NextJS](https://nextjs.org/docs/llms-full.txt), [Stripe](https://docs.stripe.com/llms.txt), [BetterAuth](https://www.better-auth.com/llms.txt)and [Zod](https://zod.dev/llms.txt) already provide this. > > With this file we can make sure that the AI model fetching it is always using the latest features and versions of postgresql. > > https://llmstxt.org/ I would recommend posting this to the pgsql-www@ mailinglist instead to get the attention of the website team. That being said, AFAICT there are no LLM providers who have publicly stated that they are using this proposed standard at all, and the author has not submitted it to IETF for a formal review in the RFC process, so it doesn't seem all that promising. -- Daniel Gustafsson