Description
🐞 Issue
autoresponses are inaccurate and have no contextual analysis to determine the need of being sent (wit.ai), logfile analysis is needed to ease aid
❗ Solution
Seperate out the autosupport into its own application, ideally not in rust
add log file parsing that tries to find issues and provide solutions alongside displaying useful information for human support
add metadata to logfiles so that client and OS versions etc can be parsed out alongside common errors and unknown exceptions (ie citra)
use wit.ai and create a dataset that is fed with data from the guild to train the model accordingly which will then spit out solutions to given problems when it identifies them
❓ Motivation
former djs staff here, seen and made these kinds of applications, current state of this is not great, rust is not quite ideal for discord development, regex based responses tend to just clutter everything and be - as they currently are - too intrusive and inaccurate, I've created wit.ai datasets before and can vouch for this to work very well if given enough data to train, which there is plenty of in the support guild
⚠ Additional context
citra log analysis https://cdn.adriancastro.dev/vWcgvpf.png
djs support responses https://cdn.adriancastro.dev/M1BLruI.png
wit.ai https://wit.ai/
224617799434108928 on discord in case you need help with the dapi ecosystem or the wit.ai datasets or just have general questions in regards to my sanity.