quick_validate.py fails on UTF-8 SKILL.md under Windows CP949

[jeonking80]

Environment

• Windows 11
• Python locale encoding: CP949

Problem

skills/.system/skill-creator/scripts/quick_validate.py reads SKILL.md with Path.read_text() and therefore uses the Windows locale encoding. A valid UTF-8 skill containing Korean text or an em dash can fail with UnicodeDecodeError.

The validator also rejects host-registered optional frontmatter fields such as argument-hint, user-invocable, homepage, repository, and author before validating their types.

Expected

• Read SKILL.md explicitly as UTF-8.
• Accept the registered optional fields.
• Validate user-invocable as a boolean.

Proposed patch

• Commit: jeonking80@4b18e14
• Compare: main...jeonking80:skills:skill-hardening-task4

Verification

On Windows, the three focused regression tests pass and the modified skill-creator validator passes.

[jeonking80]

The logged-in GitHub compare UI confirms that repository owners have disabled opening pull requests. Please use the linked fork commit/compare as the patch source if this fix is accepted: jeonking80@4b18e14

[wenqifu]

Additional current-bundled reproduction: this also fails under Windows CP1252, so the defect is not specific to CP949.

• Python: 3.13.14
• locale.getpreferredencoding(False): cp1252
• sys.flags.utf8_mode: 0
• Plain py quick_validate.py <valid-multilingual-skill>: exit 1 with UnicodeDecodeError from encodings/cp1252.py (0x81, position 4251)
• py -X utf8 quick_validate.py <same-skill>: exit 0 with Skill is valid!
• The currently bundled source still calls skill_md.read_text() without an explicit encoding.

This independently supports the issue's proposed skill_md.read_text(encoding="utf-8") correction. I did not patch the bundled system skill locally; the operating workaround remains py -X utf8.