Also tagging @rachfop since you are the owner of the repo.

this feature would be tremendously helpful for our AI agentic workflows at Sablier

thanks @smol-ninja, would you mind testing what happens if someone has not defined an explicit ID (e.g. you quickly remove some of the ids in your repo and see what the generated file name is), because if it is actually picking up the implicit ID as outlined here: https://docusaurus.io/docs/create-doc#document-id I would vote for removing the useIdForFilenames and always do that instead of that code that is figuring out the filepath itself.

would you mind testing what happens if someone has not defined an explicit ID

In that case, it picks up the file name as usual.

would vote for removing the useIdForFilenames

Are you suggesting to make it a default behaviour i.e. if id is mentioned, use that for filename always?

Are you suggesting to make it a default behaviour i.e. if id is mentioned, use that for filename always?

Yes, based on the documentation from docusaurus, it sounds to me that there is always a DocumentID (https://docusaurus.io/docs/create-doc#document-id):

Every document has a unique id. By default, a document id is the name of the document (without the extension) relative to the root docs directory.

So if we can capture that ID somehow, then we have covered all the cases in one.

So if we can capture that ID somehow, then we have covered all the cases in one

Unfortunately, I could not find if Docusaurus offers an API to read that ID. So, I have added the following logic:

1. Priority 1: use frontmatter's slug as file name (since its prioritised over doc ID)
2. Priority 2: use frontmatter's id as file name
3. Fallback to actual file name

The new tests cover all cases.

Let me know if it looks good: b28e843

@smol-ninja thank you for the PR. This looks great.

Thanks for merging the PR. When will it be released?

Can we please cut a new release? We really need this.

We also need this feature for our documentation. We're currently using docusaurus-plugin-llms@0.2.2 to generate llms.txt files for our SDK documentation.
The issue we're facing is that the current version generates URLs based on file paths instead of the slug values defined in our frontmatter. This results in incorrect URLs in the generated llms.txt file, as our documents use custom slug values to maintain clean, consistent URLs that differ from the actual file paths.
Could you please cut a new release with this feature? It would ensure that the generated URLs match the actual documentation URLs used in our Docusaurus site.
Thank you for your work on this plugin!