TasteSkill v2: Adding Art Direction and Real Assets to Coding Agents

X postLeon LinAnnouncement

Leon Lin announced TasteSkill v2, rebuilt from scratch to equip coding agents with structured art direction, real assets, and browser verification checks when generating web interfaces.

Key points

  1. TasteSkill v2 analyses project briefs to extract distinct client requirements.
  2. A script researches twenty design directions and offers three choices for the agent to select from.
  3. The agent locates real photos, maps, and other assets online and integrates them with proper attribution.
  4. Automated browser checks verify visual output during the generation process.
  5. Available via the command 'npx skills add Leonxlnx/taste-skill'.

Why it matters

A frequent drawback of AI-generated frontends is their generic aesthetic and reliance on synthetic placeholders. TasteSkill attempts to automate design taste by guiding agents through curation and real asset sourcing.

Analysis

The project tackles the aesthetic limitations of LLM-driven UI design. Rather than letting the model fall back on typical defaults, TasteSkill constrains the process into structured creative choices and real-world asset discovery. While Lin acknowledges that certain models like Argon still struggle with UI design, he has already promised a v2.1 update targeting dashboards alongside future motion design plugins.

Original post

Leon Lin

@LexnLin · 13 October 2026

We rebuilt TasteSkill from scratch And today we are introducing TasteSkill v2 v2 gives your coding agent an art direction, assets and a browser check to verify its work npx skills add Leonxlnx/taste-skill check out more examples below! t.co/coX4St1t3a

2.4k likes63 replies3.1k bookmarks145.7k views

View on X
More posts in the thread (11)
  1. how it works: - It reads a brief and writes down the things only this client has - Then a script gives it three directions out of the twenty it researched, and it picks one of them - your agent will find real photos, maps and other assets and use them (with the credits). - with
  2. Some outputs we got with simple prompts! t.co/P8pPUK4RZm
  3. btw the launch video is code as well made with Cinetic and Opus 5.5 and I'm planning to improve the skill a lot in the coming days, stay tuned :) tastecode updates coming very soon as well!
  4. repo: t.co/ByHmn0Cx0F photo credits for everything in the video: t.co/lRArmdCXkP
  5. @notjazii ty man! I'm kinda never happy *enough* with the results, a lot of improvements coming soon!!
  6. @be_arsh same lol, Im working on improving it asap if it doesnt work well with argon. cuz argon tbh is pretty bad at UI.
  7. @ziwenxu_ thanks for the support🙏🏻🙏🏻
  8. @aiwithags in this video as well? these are actually real assets used lol not some synth sounds that opus always uses. but I'm working on it!
  9. @blckgenesiz I AM, so many updates coming for tastecode, tasteskill v2, desing PLUGINS, motion design skills, a few other cool projects and more 🙏🏻🙏🏻
  10. @impulseraid i promise you that ill put out v2.1 by end of next week that makes dashboard GOOD as well otherwise ill get you a claude pro sub
  11. @aiwithags no synth music is just pretty standard for opus. and the bell sounds hahaha. I have an update coming that will tell opus to use free assets online to improve the sfx and music
Selected replies (5)
  • @notjazii @LexnLin My go to skill. You really cooked
  • @leodev @LexnLin @daradoescode time to update whichai dev?
  • @vivescene @LexnLin Was the website updated according to this skill? If so - it still looks like ai slop :(
  • @be_arsh @LexnLin Going to try this with Gemini 4 Argon right now.
  • @ziwenxu_ @LexnLin always love the taste skill

Summary generated by AI (Gemini 3.8 Flash, 15 October 2026). It may contain errors – the original source is authoritative.

Similar in content, found by the AI search.

  • X post:Adham Dannaway

    Design Skills for AI Agents: Equipping Coding Agents with UX and UI Knowledge

    Adham Dannaway highlights 'Design Skills for AI Agents', an open collection developed by Carlos Cuéllar. The repository provides structured design heuristics, WCAG 2.2 accessibility rules, and design system frameworks for tools such as Claude Code, Cursor, and Codex.

    100bookmarks5.2kviews

    AI & LLMs· Resource

  • Link:Emil Kowalski

    Emil Kowalski's Design Engineering Skill for AI Agents

    Emil Kowalski has published an installable agent skill based on his design engineering philosophy. It aims to guide AI coding assistants in building polished, detail-oriented user interfaces.

    AI & LLMs· Resource

  • X post:julio

    /awwwards: Agent Skill Generating Award-Calibre Websites from Minimal Prompts

    Developer Julio released the /awwwards agent skill, designed to build award-calibre websites from single-line prompts. Compatible with coding agents such as Claude Code, Cursor, and Codex, the tool executes an automated design and review workflow featuring browser screenshots and accessibility audits.

    1.3kbookmarks44.4kviews

    AI & LLMs· Tool

  • Link:cuellarfr

    Design Skills: Curated UI and UX Skills for AI Coding Agents

    cuellarfr/design-skills provides ten structured skill packages for developer agents including Claude Code, Cursor, Codex, and OpenCode. The library covers areas ranging from UX research and accessibility to design systems and UX writing.

    AI & LLMs· Resource

Let's talk about your project

Locations

  • Mattersburg
    Johann Nepomuk Bergerstraße 7/2/14
    7210 Mattersburg, Austria
  • Vienna
    Ungargasse 64-66/3/404
    1030 Wien, Austria

Parts of this content were created with the assistance of AI.