when reading curios using the curio capability (I didn't try other methods, so I don't know if it behaves the same there), in about half of the launches, the order in which the first two slots (slot 0 and slot 1) appear is different between (logical) client and (logical) server.

this causes a problem with ae2 / ae2wtlib, which uses the slot number to tell the client where to find an item to open a GUI. (Mari023/AE2WirelessTerminalLibrary#312)

I could reproduce that issue by putting an ae2 wireless terminal in HostileNeuralNetworks curio slot, and then using the keybinding to open the terminal.
debug crash report from the instance I used to reproduce this: https://gist.github.com/Mari023/b236e4efbc0dee4031392f64381091d6

Sorry for not getting back to you on this issue but I am quite confused about the issues as it seems the slots for a given slot type are in different order i.e. 0 -> 1 and then after relaunch, it can be 1 -> 0 which makes no sense based on how any of the Accessory API works. Frankly having a hard time reproducing any issue in production using AE2 and Accessories Slots specifically

Also, the debug report dose not seem that helpful besides giving a mod list so idk if such is intentionally the only info that can be given here.

Also, the debug report dose not seem that helpful besides giving a mod list so idk if such is intentionally the only info that can be given here.

it's use is mostly the modlist, sometimes also other system information.
lots of issue trackers require them, so I just included it by default.
I also have this log from when I triggered the bug again now, tho I don't think it is any more useful since it just shows ae2 complaining it can't find the item in the slot it expected to find it.

I am quite confused about the issues as it seems the slots for a given slot type are in different order i.e. 0 -> 1 and then after relaunch, it can be 1 -> 0 which makes no sense based on how any of the Accessory API works

It doesn't make sense to me either, but that was what I observed.
To me it makes even less sense that there is a discrepancy between the client and the integrated server at all.

It also appears normally in the UI.

I tried it a few times without Hostile Neural Networks, and couldn't reproduce it, so it is possible they are doing something weird

Edit: glancing at their code, I couldn't find anything special, they are just adding the slot using a normal curios datapack

Might be related to this issue which the code is based on meaning this may be the actual issue: TheIllusiveC4/Curios#363

Ill try using the update version to see if it fixes the issue

Sadly did not fix the issue as I thought it might ):