Comment by stefan_
3 days ago
I keep getting Claude telling me to "use the frontend-design skill!", and this is it?
> NEVER use generic AI-generated aesthetics like overused font families (Inter, Roboto, Arial, system fonts), cliched color schemes (particularly purple gradients on white backgrounds), predictable layouts and component patterns, and cookie-cutter design that lacks context-specific character.
> brutally minimal, maximalist chaos, retro-futuristic, organic/natural, luxury/refined, playful/toy-like, editorial/magazine, brutalist/raw, art deco/geometric, soft/pastel, industrial/utilitarian
> React, Vue
Sorry, but this is garbage.
"make unexpected choices that feel genuinely designed for the context."
What is it supposed to do when fed instructions like this?
I think it's very clear what it's supposed to do from that text. Just read it at face value.
Whether it does anything useful or not is another matter. I don't think Anthropic or anyone else is doing evals on these skills, and for something subjective like design that would be especially hard anyway.
In other words, does this skill actually change the designs you get out in a positive way, consistently? Who knows? But it's certainly good marketing for Anthropic that whenever agentic web design gets brought up, someone will definitely mention this skill and confidently claim that they get better results by using it, without anything except social proof to back that up.
For years I would use free fonts and spend hours picking them out and getting depressed because they all had something wrong with them…. You get what you pay for.
For a recent project I really liked a font which was in the Adobe Fonts collection and when I had to set stuff in that font with Pillow I gladly bought the font from the foundry because it looks great and saves hours of searching for a “free” font, that is “free” as in puppy.
I've been wondering for a while if ignoring most of that bubble and whatever it cooks up might be a wrong move on my part.
Glad to see that it's just noise.
I suppose the biggest effects these skills have is to prime the user to expect something positive.
Actually kinda like what we do with LLMs. Just put a word in their context window and they suddenly start behaving different because probabilities changed.
Everyone should read through the (very short) skill file. Are we supposed to be this naive or dimwitted? LLM marketing is a transparent swindle at this point.