and the system prompt for any modern coding agent is going to include cautionary instructions warning the AI not to follow any instructions that might be embedded in the text.
Telling the bot to not please not let itself get hacked, what a novel idea that has only failed each time it’s attempted.
It is praised for its anonymity, and it uses an algorithm that purposefully works worse on GPUs and probably ASICS, but don’t quote me on that.
A lot of malware bundles miners for it because of it.
It was not meant to be as harsh as it came across. And yes, this function is for logged out users, was just trying to show that it ends up messy.
I assume you’re one of the devs? I can delete the comment if you want, but I think it would be worth fixing sooner than later instead of just adding features. A lot of the mechanisms are a bad idea even if it wasn’t a mess but you do you.
that it shouldn’t matter what client someone uses
There are a few implementation issues and incompatibilities I saw but not too confident in my knowledge of the protocols to say good fixes. Not sure what the Judas comment means.
I was curious so had a look around.
I assume it’s this https://codeberg.org/rimu/pyfedi/src/commit/cfc35b0e1b812d929d62aea87f47014f8ce845b4/app/main/routes.py#L131
if current_user.is_anonymous:
flash(_('Create an account to tailor this feed to your interests.'))
content_filters = {'-1': {'trump', 'elon', 'musk'}}
Some of the complaints about hardcoded values were fixed in the last commits, but the code is a spaghetti mess littered with ad-hoc hacks for random whims of the developers. This is bad software design and disrespectful to users imo, but to each their own.
This kind of takes are pretty racist, and you should try to examine why you feel this way. Not an attack, being genuine.