interface-auditor — Claude Code skill: UX antipattern audit
Reads an interface description and flags UX antipatterns, rating severity and returning fix steps as JSON.
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a claude-codeWhat interface-auditor does
This skill turns a written screen description into a UX smell report. The taxonomy holds 8 smells split across 6 categories, among them efficiency, feedback, accessibility and consistency. Signals from the text get matched against symptoms, and each smell lands in one of two buckets: confirmed or merely suspected.
Output is JSON only, no narrative. Each finding carries its id, the symptoms that matched, and a severity of critical, warning or suggestion. Detection follows a fixed rule: two matching symptoms, or a single critical one with high confidence, and the smell counts as found. Because the shape never changes, you can parse it in a script.
Richer answers come from uxuiprinciples.com over curl: full numbered remediation recipes, time estimates and related principle codes. That path needs UXUI_API_KEY set in the environment. Skip the key and the skill falls back to its own taxonomy knowledge and marks the reply with api_enriched: false, which means no recipe steps. You get a diagnosis, not a repair plan.
What it helps with
- catch design flaws in a screen description before anyone codes it
- explain why a signup form loses users
- turn a wireframe write-up into a UX issue list
- sort blocking accessibility problems from minor ones
- feed structured audit findings into a script
Example prompts for the agent
- Audit this signup screen description for UX smells: ten fields, one
Continuebutton, errors only after submit - Check the feedback category against my empty-state write-up
- Run the smell taxonomy on this navigation description and give me JSON
How to install interface-auditor
1Skills CLI (Vercel)
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a claude-code2Manually: put the files in place
~/.claude/skills/interface-auditor/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/interface-auditor/
1Skills CLI (Vercel)
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a codex2Manually: put the files in place
.agents/skills/interface-auditor/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a cursor2Manually: put the files in place
.cursor/skills/interface-auditor/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install uxuiprinciples/agent-skills interface-auditor2Manually: put the files in place
.github/skills/interface-auditor/SKILL.md1Skills CLI (Vercel)
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a gemini-cli2Platform CLI
gemini skills install https://github.com/uxuiprinciples/agent-skills --path interface-auditorManually: put the files in place
.devin/skills/interface-auditor/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about interface-auditor
How do I install interface-auditor in Claude Code?
Run npx skills add uxuiprinciples/agent-skills --skill interface-auditor -a claude-code. The skill then sits at ~/.claude/skills/interface-auditor/SKILL.md.
What does interface-auditor do?
It maps an interface description onto the uxuiprinciples smell taxonomy and returns JSON findings with symptoms and severity. With an API key it also pulls full remediation recipes.
Do I need an API key to use it?
No, the key is optional. Without UXUI_API_KEY the skill works from its built-in knowledge of the taxonomy and flags the result as not API-enriched, so remediation steps and time estimates are left out.
