Ultra-thorough Claude Code skills for design, security, testing, and performance audits
This repository also configures its own agents. See what picky-skills tells them →
Skill Claude CodeCodex
Ultra-thorough design system and UI audit. Checks every component for visual consistency, design token usage, accessibility (WCAG 2.2 AA), and responsive design. Produces scored reports with exact file:line references. Trigger with "design audit", "accessibility audit", "UI consistency check", "picky design".
Skill Claude CodeCodex
Ultra-thorough landing page conversion audit. Scores every element against a 10-point conversion framework covering value prop, hero, social proof, CTA, copy, pricing, SEO, and mobile. Produces exact find/replace fixes. Trigger with "landing page audit", "conversion optimization", "CRO audit", "copy audit", "picky…
Skill Claude CodeCodex
Ultra-thorough performance audit with Core Web Vitals focus. Analyzes bundle size, code splitting, images, fonts, caching, API patterns, runtime performance, and N+1 queries. Produces scored reports with exact measurements and fixes. Trigger with "performance audit", "why is this slow", "Core Web Vitals", "bundle…
Skill Claude CodeCodex
Ultra-thorough security vulnerability audit. Scans every API endpoint, user input, and dependency with OWASP Top 10 mapping and CWE references. Produces scored reports with CVSS ratings and exact file:line fixes. Trigger with "security audit", "vulnerability scan", "scan for secrets", "OWASP audit", "picky security".
Skill Claude CodeCodex
Ultra-thorough black-box QA testing that behaves like a real user, not a developer. Tests forms, navigation, interactive elements, error handling, and edge cases via Chrome DevTools MCP. Produces scored friction logs with screenshots. Trigger with "test this app", "QA testing", "test as a real user", "picky tester".
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: