Skill Claude

UX Expert

ux-expert/

2 views 0 downloads

UX research and design critique for your agent: research plans, personas, storyboards, heuristic reviews and measured design forensics, with at most three priority actions.

UX Expert: Brain Lab skill for Claude Code

UX research and design critique, the reasoning BEFORE a wireframe and the judgment AFTER a build.

Why use it

Design decisions defended by taste instead of evidence.
"Analyze this design" answered with guessed hex codes and invented font sizes.
A reviewer who recites all ten heuristics and prioritizes nothing.

What it does

research, pick the method that fits the question × product stage, and mine the free evidence first.
personas, 2-4 relevance-tested personas, each carrying its provenance (grounded, or clearly labelled proto).
storyboard, one persona · one scenario · one path, with an emotion per panel.
heuristics, Nielsen H1-H10 applied selectively: critique, or design guidance handed to the designer.
review, the combined lens (usability · cognitive load/conversion · craft · a11y · AI-product) with a verdict and max 3 priority actions.
analysis, design forensics: measured hexes, type scale, spacing ladder, contrast pairs, composition metrics, tokens.

Benefits

Numbers, not opinions. Every hex, size and ratio is measured, anything inferred is labelled.
Findings you can act on today: 3-5 issues, each with its cost, and max 3 priority actions.
Know who you're designing for before the first wireframe, and whether the build served them after.
No turf wars. It critiques and routes; it never touches your design system or your code.

Install

Download this skill and copy the ux-expert folder to .claude/skills/ux-expert/ in your project, or to ~/.claude/skills/ux-expert/ to use it in every project. Claude Code finds it on its own; you can also call it with /ux-expert.

Part of the Web Dev Design Pipeline bundle. To get the whole set as a Claude Code plugin: /plugin marketplace add endorphin-ai/hasbrains-agent-kit, then /plugin install web-dev@hasbrains-agent-kit.

Source: endorphin-ai/hasbrains-agent-kit. License: MIT.

Files (29)

{
  "source": "https://github.com/tommyjepsen/awesome-ux-skills",
  "source_files": [
    "ux-research-methods.md",
    "ux-personas.md",
    "ux-storyboard.md",
    "ux-heuristics-review.md",
    "general-design-review.md",
    "design-analysis.md"
  ],
  "source_type": "external-skill-set",
  "built_at": "2026-07-29T00:00:00Z",
  "updated_at": "2026-07-31T00:00:00Z",
  "built_with": "build-brain (lean map + atomized tagged notes + nav)",
  "version": "1.1",
  "counts": { "notes": 18, "references": 12, "workflows": 6, "tags": 23, "links": 18 },
  "tokens": { "skill_md": 3400, "index_json": 2900, "references_total": 19200, "workflows_total": 6100 },
  "local_additions": [
    {
      "note": "verification-instruments",
      "origin": "learned in production use, NOT from the external source",
      "evidence": "nine probe failures on one site, each of which produced a confident wrong answer that was reported as fact before being caught"
    },
    {
      "note": "honest-rendering",
      "origin": "learned in production use, NOT from the external source",
      "evidence": "nine informational-integrity violations found on one site, none of them visual"
    }
  ],
  "adaptations": [
    "Evidence paths -> the project's git-ignored scratch folder; durable artifacts -> docs/ux/",
    "Dev target {config.dev_url}; deployed target = the per-PR / staging preview only, never production",
    "Breakpoints read from {config.frontend_runtime} (a common ladder is 375/768/1024/1440)",
    "Token idiom = whatever token source the project declares; never invent or import a second one",
    "Craft rules 1/2/7 (gradients, glow, pure black/white) are overridden by the project's own approved design system",
    "Browser verification = drive a real browser (agent-browser CLI / Playwright / browser MCP) and click every control",
    "Visual-drift findings route to designer-frontend-contract; defects become docs/bugs/ bug + RCA files"
  ]
}