LLM documentation

[anyb1s]

Can the docs be somehome available to AI editors to use, ie Cursor? Dont think it crawls the pages where requires authentication

[Hesesses]

+1

[danrocha]

Yes, it would be great to have either the docs as a simple .md or collection of .md files available to download or on a single URL available to scrape.

[kriiv]

This would be a massive productivity booster.

[scarselli]

+1. I've even tried scraping the docs and because the code is placed within <textarea>, the markdown conversion gets all messed up and I've ultimately been unable to scrape the docs into a markdown file successfully.

This would be a game-changer feature.

[joshhanley]

@wishborn sorry I've had to delete your comment. While we appreciate the effort, Flux is a paid product and so redistribution of Flux (and related information) is against the licence agreement, hence why I have deleted your comment. We understand that the limitation of needing to be logged in sucks as LLMs can't fully read the docs. But we also have our reasons why users need to be logged in at this stage.

[wishborn]

Everything that was in those docs is publicly available already. All I did was make it easier for the AI to read it.

[christopherarter]

@wishborn sorry I've had to delete your comment. While we appreciate the effort, Flux is a paid product and so redistribution of Flux (and related information) is against the licence agreement, hence why I have deleted your comment. We understand that the limitation of needing to be logged in sucks as LLMs can't fully read the docs. But we also have our reasons why users need to be logged in at this stage.

@joshhanley I can empathize for sure. Perhaps an LLM friendly text doc behind an auth endpoint for paid users could be a value add and low hanging fruit e.g. /api/docs/llm?api_token=<xyz> and easily cacheable so ideally wouldn't incur much resources on Flux side either?

I'd even buy this as a plugin / upgrade to my existing unlimited license.

Just spitballing! 👍

[joshhanley]

@wishborn yep, you're more than welcome to do that for your own uses. I'm just asking that you don't share that for the moment. As I said, I get it, I understand. I use LLMs everyday too. But I'm only working off what our policy currently is. We will be reviewing it, but everything takes time.

@christopherarter great idea for sure! Our docs aren't formatted in a way that makes it easy to generate at this stage 😅 the reference docs we've just added are step in that direction though 😁

[wishborn]

We are going to see more and more tool and API providers attaching MCP servers to their docs. This would go a long way towards automated development.

…

On Mon, Mar 17, 2025 at 12:05 PM Chris Arter ***@***.***> wrote: @wishborn <https://github.com/wishborn> sorry I've had to delete your comment. While we appreciate the effort, Flux is a paid product and so redistribution of Flux (and related information) is against the licence agreement, hence why I have deleted your comment. We understand that the limitation of needing to be logged in sucks as LLMs can't fully read the docs. But we also have our reasons why users need to be logged in at this stage. @joshhanley <https://github.com/joshhanley> I can empathize for sure. Perhaps an LLM friendly text doc behind an auth endpoint could be a value add and low hanging fruit e.g. /api/docs/llm?api_token=<xyz> and easily cacheable so ideally wouldn't incur much resources on Flux side either? I'd even buy this as a plugin / upgrade to my existing unlimited license. Just spitballing! 👍 — Reply to this email directly, view it on GitHub <#925 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADQVBW2C53TYRWNEKKCGVL2U36FJAVCNFSM6AAAAABUETXPNKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTENJSHAYTQMQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Cheers, -=Wish

[wishborn]

yeah no doubt, I wasn't trying to push any particular point. Just mentioned that as a suggestion on how to serve devs that use this product (and all products), better. You had mentioned exposing an llm api endpoint, so I mentioned the MCP server since that is exactly what the purpose of an MCP server is for, (exposing systems data/knowledge to llms). P.S. Not assuming you don't know this, I am just being very descriptive for anyone else that might read this and isn't familiar with the terms.

…

On Mon, Mar 17, 2025 at 4:02 PM Josh Hanley ***@***.***> wrote: @wishborn <https://github.com/wishborn> yep, you're more than welcome to do that for your own uses. I'm just asking that you don't share that for the moment. As I said, I get it, I understand. I use LLMs everyday too. But I'm only working off what our policy currently is. We will be reviewing it, but everything takes time. @christopherarter <https://github.com/christopherarter> great idea for sure! Our docs aren't formatted in a way that makes it easy to generate at this stage 😅 the reference docs we've just added are step in that direction though 😁 — Reply to this email directly, view it on GitHub <#925 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADQVBWSCJFJWMFEHL73ADL2U42AFAVCNFSM6AAAAABUETXPNKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTENJTGAZTSMI> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Cheers, -=Wish