Detect the language of the user's request and generate the entire report in that language:
If the user writes in Chinese (e.g., "检查我的skill", "给这个skill打分", "这个skill符合规范吗"): generate the full report in Chinese, including all section headings, findings, improvement suggestions, and the quick fix checklist.
If the user writes in English (e.g., "check my skill", "score this skill", "does this follow the spec"): generate the full report in English.
If the language is ambiguous or mixed: default to English, but add a note at the top of the report: "Note: Report generated in English. Reply in Chinese if you'd prefer a Chinese report."
Apply this language choice consistently throughout the entire report — do not mix languages within a single report.
What This Skill Does
Given a path to a skill directory or SKILL.md file, you will:
Read every file in the skill directory (SKILL.md, scripts/, references/, assets/, and any other files)
Check each file and each line against the official spec rules
Score the skill across six dimensions (total 100 points)
Generate a detailed Markdown report (in the user's language) with findings and prioritized improvement suggestions
Validation Process
Step 1: Discover the Skill Structure
First, understand what you're looking at. The user may provide:
A path to a skill directory (e.g., /path/to/my-skill/)
A path to a SKILL.md file (e.g., /path/to/my-skill/SKILL.md)
A GitHub URL (e.g., https://github.com/user/repo/tree/main/skills/my-skill)
A description of their skill in the conversation
For local paths: list all files in the directory recursively. For GitHub URLs: fetch the directory listing and then each file. For conversation-provided content: work with what's given.