While i successfully set up the server using uniskin api model and client can request the texture. There are some problem i came across which is not explained in docs or wiki or repo:

• what is the condition for cache to be invalid? last_updated parameter from uniskin server api response seems to be ignored by the mod
• anyway for server to force the mod to delete cache?
• how to disable cache entirely?
• condition for elytra to be loaded/requested using api other than legacy

i think i have to skim the source to understand how it works. for now...

Also, I have set up the skin server. Where's the config for the client to point to the server?

Also, I have set up the skin server. Where's the config for the client to point to the server?

I did it this way 😄
https://pastebin.com/Dda7SdYL

@PanoptesDreams UniSkinAPI has English document, you can use this API.
https://github.com/RecursiveG/UniSkinServer/blob/master/doc/UniSkinAPI_en.md
Then add this in CustomSkinLoader's config loadlist:

{
  "name":"example",
  "type":"uniskinapi",
  "root":"https://www.example.com/"
}

@xtrial

• We judge the cache like your browser. If Expires or Cache-Control specified, we will follow it. If we find the cache expires, we will make a request. Otherwise, we will use local cache.
  Code about expire judge: https://github.com/xfl03/MCCustomSkinLoader/blob/csl-14/Common/source/customskinloader/utils/HttpRequestUtil.java#L237-L249

• You can set short Expires or Cache-Control BEFORE PROFILE LOADED, we will try to request profile online everytime.
  Code about cache check: https://github.com/xfl03/MCCustomSkinLoader/blob/csl-14/Common/source/customskinloader/utils/HttpRequestUtil.java#L95-L115

• In UniSkinAPI document, we can see a model named elytron, it means elytra.
  Commit: a6666b5

Thanks for your question.

@xfl03 while i have found the answer, i appreciate the clear reply nonetheless. I think this will answer question about cache in the future.