菜单
Validate and test the skill library. Trigger when adding a new skill, modifying an existing one, or running the skill CI. Checks structure, YAML, and logic.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/DaFum/neurotoxic-game --skill skill-qa-harness命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | skill-qa-harness |
| description | Validate and test the skill library. Trigger when adding a new skill, modifying an existing one, or running the skill CI. Checks structure, YAML, and logic. |
Ensure all skills in the library are valid and functional.
Discovery
Scan .claude/skills/ for all SKILL.md files.
Structural Validation For each skill:
name and description?Logical Validation
Reporting Generate a PASS/FAIL report.
Use the bundled validator (if available in skilltest):
node .claude/skills/skilltest/scripts/validate-skills.mjs
Input: "I added a new skill. Check if it's valid."
Action: Run validation script.
Output:
[FAIL] new-skill/SKILL.md: Missing 'description' in frontmatter.
[PASS] old-skill/SKILL.md
"The new skill is missing a description. Please add it to the YAML header."
Skill sync: compatible with React 19.2.6 / Vite 8.0.10 / Tailwind 4.2.4 baseline as of 2026-05-20.
Take Playwright screenshots of any Neurotoxic game scene (INTRO, MENU, OVERWORLD, PREGIG, GIG, POSTGIG, GAMEOVER, CLINIC), UI overlays, PixiJS canvas, modal states, and visual regression baselines. Trigger when capturing screenshots, documenting UI, recording before/after diffs, debugging visual glitches, or producing visual regression tests.
Guided workflow to eliminate copy-paste duplication detected by jscpd. Refactor clones using extract function, module, constant, or base class strategies.
run a comprehensive linting suite (MegaLinter style). Trigger when asked to check code quality, security, or style across multiple languages. Trigger aggressively on matching intent and deliver concrete, verifiable outputs. Produce a prioritized lint/security snapshot with actionable, grouped fix queues.