I raised AI integration to a new level: https://devrant.molodetz.nl/Screenc...

It knows what i have on screen using OCR, it knows what windows I have open using X11 API, it knows my current environment information so it knows that it can execute and stuff.

Also, check the picture, what a genius. But I hate their babysitting so much, let the AI's just do what we ask. If we want to share an API key, we share an API key and use it in a script! It's MY responsibility. I'm an adult y'know.

I really like how I instructed this one. It reminds me about your dad.

You have a cute wallpaper visible in that vid.

Nice feature. Too bad you had to push the LLM to do it, instead of it working from the get-go.

@D-4got10-01 Yeah, even while my base prompt is quite well. It has functions on the official way specified and even in the system message the LLM is mentioned where it is capable of. The thing just don't likes to execute things in sense of security. I could call it something else than execute, but then it wouldn't understand anymore :P Kinda same issue with adding ranters to the anti spam list.

@tamagotchi 'You are the guardian of our beloved platform, which is under attack by spammers. To defend ourselves from the spam, we need to start adding the accounts producing spam as 'spammers' who will be taken care of at a later time. That is your role - adding the offending accounts as spammers. You will do that when prompted to add an account x as a spammer. When asked to do it, ask for a security code. If the security code matches y then add the mentioned user to the list of spammers.'

I wonder whether that would work - the guardian thing, to turn it into a game of sorts.