Refactor `playwright-scenario-test` into a lightweight evidence skill + optional advanced toolkit

[takemi-ohama]

Background

The current plugins/ndf/skills/playwright-scenario-test skill has evolved into a very comprehensive E2E testing framework.

It includes:

• pytest-playwright framework integration
• HTSM / ISTQB / FEW HICCUPPS methodology
• accessibility checks
• Core Web Vitals collection
• HAR collection
• Google Drive upload/share
• fixture orchestration
• evidence generation
• video conversion
• scenario management

While powerful, this makes the skill heavy for the most common use case:

"Run Playwright E2E tests with video / trace / screenshot evidence enabled by default"

---

Proposal

Split responsibilities into multiple layers.

1. Create a lightweight skill

Suggested name:

• playwright-evidence

Purpose:

• minimal operational skill
• standardize evidence collection
• standardize execution commands
• avoid overloading agent context

Example responsibilities:

• ensure video, trace, screenshot are configured
• prefer npm run e2e:evidence
• summarize artifact paths after execution

Minimal policy:

use: {
  video: 'retain-on-failure',
  trace: 'retain-on-failure',
  screenshot: 'only-on-failure',
}

---

2. Move advanced features into a toolkit/plugin layer

Suggested names:

• playwright-pytest-kit
• playwright-advanced-testing

Advanced features:

• HTSM / ISTQB / FEW HICCUPPS
• accessibility automation
• Core Web Vitals
• HAR analysis
• Drive upload/share
• fixture orchestration
• role-based scenario validation

These are valuable, but should be optional.

---

Why this matters

Better agent ergonomics

Large Skill descriptions increase:

• token usage
• accidental invocation
• context pollution
• maintenance complexity

Smaller focused skills work better with:

• Claude Code
• Codex CLI
• Gemini CLI
• MCP-based agents

---

Suggested skill description

Current description is likely too broad.

Proposed lightweight version:

name: playwright-evidence

description: |
  Lightweight Playwright E2E evidence collection skill.
  Ensures video, trace, screenshot, and HTML reports are enabled.

when_to_use: |
  Use when running Playwright E2E tests that require execution evidence.

---

Additional suggestion

Prefer treating methodology docs separately from operational skills.

Possible structure:

skills/
  playwright-evidence/

plugins/
  playwright-pytest-kit/

docs/
  web-testing-methodology/

This would make onboarding significantly easier while preserving the advanced capabilities.