Add Skill for Claude.ai

[KingLouis11]

So you can import Skills into Claude.ai
and i thought that it would be great to make compact version to Import into the Website.

Thanks

[mrgoonie]

Summary: This is a valid product/packaging request, but it needs a narrow Claude.ai import spec before implementation.

Decision: accepted / needs design

Evidence:

• The repo currently ships Claude Code/plugin and CLI-oriented skill folders (.claude/skills/..., .claude-plugin/plugin.json, and cli/assets/...).
• I did not find an existing compact Claude.ai import package/template in the repo, nor a duplicate open issue/PR for this exact Claude.ai compact import request.
• The current skill is intentionally large: it includes datasets, scripts, and generated platform assets, so simply copying the full repo into Claude.ai would likely be noisy and hard to maintain.

Next step: define a compact Claude.ai artifact separately from the CLI/plugin install path. A good first scope would be one importable SKILL.md/Project Instructions-style bundle that keeps the core UI/UX guidance and links to supported datasets, without bundling generated mirrors or helper scripts that Claude.ai cannot execute.

[clark-cant]

Maintainer recheck (2026-06-26): Issue #238 remains valid as a Claude.ai import/packaging request. Current main still ships Claude Code/plugin and CLI-oriented skill folders with no dedicated Claude.ai compact import package.

Useful brainstorm / scope framing:

Underlying problem: Users want to import this skill into Claude.ai (web interface), but the current skill package is too large and includes datasets, scripts, and generated platform assets that Claude.ai cannot execute or efficiently handle. The full repo contains 50+ styles, 161 color palettes, 57 font pairings, 161 product types, 99 UX guidelines, and 25 chart types across 10 stacks — valuable but overwhelming for Claude.ai's context window and import constraints.

Tradeoff: A full copy preserves all guidance but creates noise and maintenance burden in Claude.ai. A compact version requires deciding what's essential vs nice-to-have, and risks being too stripped down to be useful. The middle path is a curated subset that covers the 80% use case (common styles, core UX principles, popular stacks) with links to full datasets for edge cases.

Recommended narrow scope:

1. Create a compact Claude.ai import package (target: <100KB) containing:
   • Condensed SKILL.md with core UI/UX principles, top 10 styles, essential color/typography guidance
   • Key stack guidance for React, Next.js, Vue, HTML/CSS (most common Claude.ai use cases)
   • UX checklist for common project types (landing page, dashboard, e-commerce)
   • Links to full datasets in the GitHub repo for users who need deeper reference
2. Document the Claude.ai import process (copy/paste vs file upload vs Project Instructions)
3. Keep the compact package in sync with source via a generation script or manual update checklist

Risk to avoid: Creating a 'lite' version that's too generic to differentiate from basic web design advice, or duplicating maintenance effort by manually curating two separate skill packages without automation.

Concrete acceptance evidence needed:

• A working Claude.ai import package (single file or small folder)
• Documentation showing how to import it into Claude.ai (with screenshots or step-by-step)
• Size comparison: compact package vs full repo (show significant reduction)
• Test case: import into Claude.ai, ask it to design a landing page, verify it uses the skill's guidance
• Maintenance plan: how to keep compact package in sync when source data updates

Related work: Issue #164 (WordPress) and #366 (Amazon Q) are similar platform-support requests. PR #302 (Codex package) shows the pattern for platform-specific skill packaging.

Maintainer recommendation: This is a high-value request because Claude.ai is a major user entry point. A good first PR would create the compact package and test it with 3-5 common design prompts to prove it works. If no one picks this up in 30 days, it may be reclassified as long-term backlog.

Posted by github-maintain cron at 2026-06-26T19:25:00Z