Agent Skill
2/7/2026

think-through

A Socratic interview skill for thinking through technical ideas like apps, products, tools, and projects. Use this when the user says "think through [idea]", "help me think about [app/product]", "let's explore [project idea]", or wants to iterate on a technical concept before building. Asks probing questions about the problem, target users, market, technical approach, tradeoffs, and viability. Continues until the idea is well-explored, then produces a written summary and proposes directions.

N
neonwatty
0GitHub Stars
1Views
npx skills add neonwatty/claude-skills

SKILL.md

Namethink-through
DescriptionA Socratic interview skill for thinking through technical ideas like apps, products, tools, and projects. Use this when the user says "think through [idea]", "help me think about [app/product]", "let's explore [project idea]", or wants to iterate on a technical concept before building. Asks probing questions about the problem, target users, market, technical approach, tradeoffs, and viability. Continues until the idea is well-explored, then produces a written summary and proposes directions.

Claude QA Skills

QA testing pipeline for Claude Code — generate user workflows, execute them with browser/iOS/mobile/multi-user automation, translate to Playwright E2E tests, and run them in CI.

Installation

# Add the marketplace
claude plugin marketplace add neonwatty/claude-qa-skills

# Install to your project
claude plugin install qa-skills@neonwatty-qa

The Pipeline

Stage 1: Generate Workflows    → workflows/<platform>-workflows.md
Stage 2: Execute Interactively → screenshots, issue reports
Stage 3: Translate to Playwright → e2e/<platform>-workflows.spec.ts
Stage 4: Run Playwright Tests   → test results, reports
  1. Generate — Explore your codebase and create comprehensive user workflow documents
  2. Execute — Run each workflow step-by-step with browser automation, capture screenshots, generate HTML reports
  3. Translate — Convert refined workflows into Playwright E2E spec files
  4. Run — Execute Playwright tests, report results, optionally auto-fix failures

Each stage works across four platforms: desktop browser, iOS Safari, mobile browser (Chromium), and multi-user concurrent sessions.

Skills

Browser (Desktop) — 3 skills

SkillTriggerStageDescription
browser-workflow-generator"generate browser workflows"1Explores codebase, discovers all routes and interactions, creates numbered workflow docs
browser-workflow-executor"run browser workflows"2Executes workflows via Claude-in-Chrome MCP, captures before/after screenshots, generates HTML reports
browser-workflow-to-playwright"convert workflows to playwright"3Translates workflow markdown into Playwright E2E tests for CI

iOS (Mobile Safari) — 3 skills

SkillTriggerStageDescription
ios-workflow-generator"generate ios workflows"1Explores web app for iOS Safari, creates mobile-specific workflow docs
ios-workflow-executor"run ios workflows"2Executes workflows in iOS Simulator Safari, captures screenshots, generates HTML reports
ios-workflow-to-playwright"convert ios workflows to playwright"3Translates iOS workflows into Playwright tests using WebKit with mobile viewport

Mobile Browser (Chromium) — 3 skills

SkillTriggerStageDescription
mobile-browser-workflow-generator"generate mobile browser workflows"1Explores codebase, creates mobile workflow docs with iOS HIG focus
mobile-browser-workflow-executor"run mobile browser workflows"2Executes workflows in Playwright mobile viewport (393x852)
mobile-browser-workflow-to-playwright"convert mobile workflows to playwright"3Converts mobile workflows to Chromium mobile CI tests

Multi-User (Concurrent Sessions) — 3 skills

SkillTriggerStageDescription
multi-user-workflow-generator"generate multi-user workflows"1Discovers multi-user flows (auth, real-time sync, cross-user features) using parallel exploration agents
multi-user-workflow-executor"run multi-user workflows"2Executes workflows with Chrome MCP (User A) + Playwright MCP (User B), tracks sync timing
multi-user-workflow-to-playwright"convert multi-user workflows to playwright"3Translates to Playwright specs with multiple browser contexts per persona

Shared — 2 skills

SkillTriggerDescription
playwright-executor"run playwright tests"Runs e2e/*.spec.ts files, reports pass/fail/skip, optionally auto-fixes failures
mobile-ux-ci"add mobile ux checks"Generates Playwright tests that detect iOS/mobile UX anti-patterns (hamburger menus, small touch targets, FABs)

Workflow

A typical QA cycle looks like:

# Desktop browser testing
"generate browser workflows"
"run browser workflows"
"convert workflows to playwright"
"run playwright tests browser"

# iOS testing
"generate ios workflows"
"run ios workflows"
"convert ios workflows to playwright"
"run playwright tests ios"

# Multi-user testing
"generate multi-user workflows"
"run multi-user workflows"
"convert multi-user workflows to playwright"
"run playwright tests multi-user"

# Run all tests
"run playwright tests"

# Run and auto-fix failures
"run playwright tests --fix"

Requirements

  • Browser skills: Claude-in-Chrome MCP
  • iOS skills: iOS Simulator MCP
  • Mobile browser skills: Playwright MCP (primary), Claude-in-Chrome MCP (alternative)
  • Multi-user skills: Claude-in-Chrome MCP + Playwright MCP (dual-browser architecture)
  • Playwright executor: Playwright installed (npx playwright install)

Local Development

# Load local version instead of cached plugin
claude --plugin-dir /path/to/claude-qa-skills

Don't install the plugin in projects where you're actively developing it — the installed version is cached and won't reflect your local changes until pushed.

Related Plugins

Skills Info
Original Name:think-throughAuthor:neonwatty