radix

I took a look at your radix skill and wanted to share some thoughts.

Links:

• GitHub
• Your skill in the SkillzWave marketplace

The TL;DR

You're at 80/100, solidly in B territory - good fundamentals with some room to polish. This is based on Anthropic's skill best practices rubric across five pillars. Your strongest area is Ease of Use (22/25) with comprehensive triggers and clear metadata. The weakest spot is Progressive Disclosure Architecture (21/30) - your 312-line single file needs some structure work.

What's Working Well

• Triggers are chef's kiss - You've got specific component names (Dialog, Dropdown, Select), pattern intents like "create modal", and even artifact-level triggers. That's exactly what makes skills discoverable.
• Code examples are solid - The Dialog, Dropdown, and Select examples show real patterns like controlled state and the asChild prop. Developers can actually use these immediately.
• Clear constraints - You're explicit about what Radix provides (accessibility, unstyled components) and what it doesn't handle (styling decisions), which prevents confusion.

The Big One: Organize Into Multiple Files

Severity: High (+5 points potential)

Your skill file is doing too much in one place. At 312 lines, you've got MCP query patterns, three full code examples, research instructions, and success criteria all mixed together. This hurts your Progressive Disclosure Architecture score because it forces users to wade through everything to find what they need.

The fix: Create a references/ directory with:

• references/mcp-queries.md - MCP patterns and library IDs
• references/code-examples.md - The three component examples
• references/research-patterns.md - Your research section

Keep your main SKILL.md focused on quick_start → patterns → constraints → anti_patterns → success_criteria. This reduces cognitive load and token overhead.

Other Things Worth Fixing

1. Add a Table of Contents - At 312 lines, readers need signposts. Add a ## Contents section listing your major sections. Quick +2 points.

2. Slim down code examples - Full Tailwind class strings like "fixed top-1/2 left-1/2 -translate-x-1/2..." repeat across examples. Use className="{/* positioning */}" comments or show one full example, then minimal templates. Saves tokens and reads cleaner.

3. Watch the voice - Minor thing: "that you style with Tailwind" should be "styled with Tailwind". Keep it imperative throughout.

4. Move success_criteria earlier - Currently at the very end (line 294). Bring it closer to your quick_start so users know what "done" looks like upfront.

Quick Wins

Most impactful first:

• Split into reference files (+5) - biggest bang for buck
• Add table of contents (+2) - helps navigation
• Consolidate code examples (+1) - reduces verbosity
• Minor voice fixes (+1) - polish

You've got solid bones here. These structural tweaks get you from 80 → 89 territory pretty easily.

Checkout your skill here: SkillzWave.ai | SpillWave We have an agentic skill installer that install skills in 14+ coding agent platforms. Check out this guide on how to improve your agentic skills.