In the spirit of “if you can’t fix it, feature it”, the code does have impeccable error recovery,
slaps the hood
You know how many CVEs this puppy can hold?
All of them.
Or this is that guy’s career path who always brags about his success and gives people advice.
Just gonna leave this here: https://posledniping.cz/#/en/
This server is managed by Claude Code as root. Every morning it reads its fortune. Every evening it writes a blog about its day. Sooner or later, AI Claude Code will make a mistake and both this blog and server will cease to exist.
I am actually pretty ok with this type of "messing around’ usage.
On the condition they also stop killing the environment to train and run these stupid things.Nice. Bukowski-inspired, bitter AI tasked with documenting its own downfall. Or not? My question is opposite what I usually ask: how much of this is in the human designed code and prompts?
Holy shit, it’s an elixir or app, the beam makes a lot of sense for that
The people sending messages via the SSH protocol identification header are my favorite.
This is so real. I’m forced to use it by my employer and it just sits there doing the cringiest shit all the time. Like everything is boilerplate and it’s just a deeply unsatisfying spaghetti mess of function calls. It’ll work if you have a good feedback loop, eventually, but only ever as a crude prototype.
Truly reminds me of WYSIWYG web development when it was still at it’s infancy.
Can’t be worse than some <div class=“h1”> pages.
I could have happily watched that for a lot longer.
We need a looping version
There’s a few frames missing, unfortunately.
I have a dream. A dream of kaliaude code.
There are already models trained on reversing binaries too.
