Imported from opendefender/OpenRisk (.claude/skills/audit-ux/SKILL.md). Install upstream with npx skills add opendefender/OpenRisk --skill audit-ux. Copyright stays with the author.

UX audit: $ARGUMENTS

Name: audit-ux
Author: opendefender

qa-automation runs the automated pass, ux-designer and art-director the judgement pass.

Automated (Playwright MCP)

Walk the flow as each of the four personas.
Capture at 1440px, 768px and 414px.
axe-core — every violation with severity and selector.
Keyboard-only traversal: focus always visible · nothing unreachable by Tab · Escape closes every overlay · focus returns to the trigger.
LCP, INP, CLS on a throttled mid-range mobile profile.

Judgement

Score the 20-criterion grid. Every score under 4 gets a concrete fix naming the file.

Output

Table: criterion · score · evidence · fix · owning agent. Then UX GATE: PASS or UX GATE: FAIL — <n> blockers, and an issue opened for each blocker.

State honestly what could not be proven headlessly. Never assert an interaction works because the build is green.

{ "ocm": "1", "id": "opendefender-openrisk-audit-ux", "kind": "skill", "name": "audit-ux", "description": "Run the OpenRisk UX audit protocol on a screen or flow — four personas, keyboard traversal, axe-core, Core Web Vitals and the 20-criterion grid. Use before closing a milestone with UI changes.", "publisher": "opendefender", "version": "1.0.0", "capabilities": { "domains": [ "general" ], "tags": [ "skill-md", "github" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "Run the OpenRisk UX audit protocol on a screen or flow — four personas, keyboard traversal, axe-core, Core Web Vitals and the 20-criterion grid. Use before closing a milestone with UI changes." ], "primary": false, "metadata": { "source": { "provider": "github", "repository": "https://github.com/opendefender/OpenRisk", "path": ".claude/skills/audit-ux/SKILL.md", "ref": "9007a6a9ea864c186f024faee12487e939603387", "url": "https://github.com/opendefender/OpenRisk/blob/9007a6a9ea864c186f024faee12487e939603387/.claude/skills/audit-ux/SKILL.md", "key": "opendefender/OpenRisk/.claude/skills/audit-ux/SKILL.md" } }, "instructions": "# UX audit: $ARGUMENTS\n\n`qa-automation` runs the automated pass, `ux-designer` and `art-director` the\njudgement pass.\n\n## Automated (Playwright MCP)\n\n1. Walk the flow as each of the four personas.\n2. Capture at 1440px, 768px and 414px.\n3. axe-core — every violation with severity and selector.\n4. Keyboard-only traversal: focus always visible · nothing unreachable by Tab ·\n Escape closes every overlay · focus returns to the trigger.\n5. LCP, INP, CLS on a throttled mid-range mobile profile.\n\n## Judgement\n\nScore the 20-criterion grid. Every score under 4 gets a concrete fix naming\nthe file.\n\n## ", "cost": { "context_tokens": 218 } }

audit-ux

About

UX audit: $ARGUMENTS

Automated (Playwright MCP)

Judgement

Output

Use it

Manifest

Reviews