@Leonxlnx I work on skill tooling, so the interesting part of Taste Skill to me is the way it turns taste into portable agent behavior instead of another design checklist. The v2 default's dials and the split between redesign, image-to-code, brandkit, and imagegen skills give visitors several concrete ways to test the system on their own frontend problem.
I set up a hosted entry point for the suite so someone landing here can try it before installing the repo:
You can also open the demo and run it with a brief, screenshots, or a codebase. That path fits the repo's existing intent: readers get an output-first way to test Taste Skill on real material, and those sessions leave usage records you can review when sharpening SKILL.md examples, defaults, and the v2 path.
The mainstream version of design prompts usually stops at "make it polished." Taste Skill is more opinionated than that: it names layout variance, motion, density, anti-slop rules, and separate workflows for generated references versus implementation. A hosted run is useful because it shows which of those entry points people actually choose.
For the usage-record review link or routing the demo to one specific skill first, email shesonglin@tinkerland.ai or ping WeChat crook-1184.
Feel free to close if this is not relevant.
@Leonxlnx I work on skill tooling, so the interesting part of Taste Skill to me is the way it turns taste into portable agent behavior instead of another design checklist. The v2 default's dials and the split between redesign, image-to-code, brandkit, and imagegen skills give visitors several concrete ways to test the system on their own frontend problem.
I set up a hosted entry point for the suite so someone landing here can try it before installing the repo:
You can also open the demo and run it with a brief, screenshots, or a codebase. That path fits the repo's existing intent: readers get an output-first way to test Taste Skill on real material, and those sessions leave usage records you can review when sharpening SKILL.md examples, defaults, and the v2 path.
The mainstream version of design prompts usually stops at "make it polished." Taste Skill is more opinionated than that: it names layout variance, motion, density, anti-slop rules, and separate workflows for generated references versus implementation. A hosted run is useful because it shows which of those entry points people actually choose.
For the usage-record review link or routing the demo to one specific skill first, email shesonglin@tinkerland.ai or ping WeChat crook-1184.
Feel free to close if this is not relevant.