La interfaz está localizada; las descripciones y documentos de las Skills se traducen automáticamente con IA a su idioma. El código, los comandos, rutas y términos técnicos se mantienen en el original, y prevalece el documento del autor.
Explorar Skills
Diseño y creatividad · SKILL.md

software-ux-research

software-ux-research

De-risk product decisions with rigorous UX research. Deep reference guides for survey design, research repository/ops, and usability testing (ISO 9241-grounded), plus JTBD.

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentación de la Skill

GitHub

name: software-ux-research description: | De-risk product decisions with rigorous UX research. Deep reference guides for survey design, research repository/ops, and usability testing (ISO 9241-grounded), plus JTBD. triggers:

  • "software UX research"
  • "UX research ops"
  • "usability test plan"
  • "survey design methodology"
  • "research repository setup" das: type: skill category: design-research upstream: "https://github.com/vasilyu1983/AI-Agents-public" upstream_path: "frameworks/shared-skills/skills/software-ux-research/SKILL.md" version: latest

software-ux-research

Catalogue stub — full skill: vasilyu1983/AI-Agents-public

Decision tree

  1. Is the full skill already installed?

    • Global: grep -q "^das:" ~/.agents/skills/software-ux-research/SKILL.md 2>/dev/null && echo "pointer" || echo "installed"
    • Project: grep -q "^das:" .agents/skills/software-ux-research/SKILL.md 2>/dev/null && echo "pointer" || echo "installed"
    • No das: block → full skill installed, invoke and proceed
    • das: block present → go to step 2
  2. Detect scope, then install:

    [ -e ~/.agents/skills/software-ux-research ] && echo "global" || echo "project"
    

    Global:

    npx skills add vasilyu1983/AI-Agents-public --skill software-ux-research -g -y
    

    Project:

    npx skills add vasilyu1983/AI-Agents-public --skill software-ux-research -y
    

    Claude Code: send either command as a chat message starting with ! to run it without leaving the conversation.

What it does

A research-ops powerhouse (~15–19KB per topic): survey-design (Likert/NPS/SUS, sample-size math with finite-population correction, bias rewrites, full stats), research-repository management (governance, taxonomy ownership, decay/freshness, GDPR/PII classification, adoption ROI), and usability testing (think-aloud protocols, SEQ/ASQ/SUS, severity matrices, reporting). Web research picked this over Owl’s thinner research skills.

When NOT to use

  • Lightweight single-method help → may be heavier than needed
  • AI-moderated/synthetic interviews → use user-research-cookiy
Skills guardadas