feat(skills): Add icon-finder skill#12
feat(skills): Add icon-finder skill#12bekah-stephens wants to merge 1 commit intopatternfly:mainfrom
Conversation
bekah-stephens
changed the title
| - **Linux:** `xdg-open icon-finder-preview.html` | ||
|
|
||
| ### Cursor | ||
|
|
There was a problem hiding this comment.
This section has separate instructions for Cursor vs Claude Code. Skills should be tool-agnostic per the contributing guidelines — a single line like "Open icon-finder-preview.html in the user's default browser" should be enough and lets each tool figure out the right command.
| --- | ||
|
|
||
| # Icon Finder (Red Hat Icons) | ||
|
|
There was a problem hiding this comment.
This searches Red Hat Design System icons (@rhds/icons), not PatternFly React components, so pf-react isn't the right home.
Would something like a pf-design-system plugin make sense to you? That could cover design system resources — icons, illustrations, typography, visual assets — and would be distinct from pf-design-tokens (CSS variable auditing) and pf-figma (Figma files). Happy to help set that up if it sounds right.
| @@ -0,0 +1 @@ | |||
| icon-finder-preview.html | |||
There was a problem hiding this comment.
The generated icon-finder-preview.html gets written to the user's workspace, not to this plugin directory — so this .gitignore wouldn't actually do anything. Safe to remove.
2 participants
Summary
This skill closes the loop on icon discovery by allowing your AI agent to search the @rhds/icons ui set (~542 icons) by described use case. It interprets natural-language prompts like "find me an icon for cloud deployment," derives search terms, fetches the live icon list from the Red Hat Icons demo site, and returns matched icons in both a markdown table and a visual HTML preview styled with PatternFly 6. Each card in the preview displays outline and fill variants side-by-side with individually clickable links to the source SVGs on GitHub, along with software-UI-specific context hints to help users evaluate which icon best fits their use case. If the initial results aren't sufficient, a deep search fallback expands to synonyms, related concepts, and domain-specific terminology to surface icons the user may not have thought to search for.
Test plan