Uth Utf8 Guard

UTH UTF-8 Guard

Purpose

Use this skill as a document-encoding guard for UTH governance docs. It prevents Chinese Markdown and governance text from being written back with broken encoding or obvious mojibake.

Trigger

Use before and after modifying:

• AGENTS.md
• root README.md
• docs/**/*.md
• task-package Markdown under docs/work/
• governed templates or manuals
• skills/**/*.md only when the user explicitly invoked skill-creator
• skills/uth-sp-*/**/*.md only during explicit UTH-SP maintenance

Do not use for normal code files, generated files, binary files, lock files, build outputs, or unrelated text unless the user asks.

Guard Protocol

Before writing:

1. If the file exists, read it as bytes and confirm it decodes as UTF-8.
2. Note whether the file already has unusual encoding, replacement characters, or mojibake markers.
3. Do not normalize unrelated line endings or rewrite unrelated content.

After writing:

1. Read the file as UTF-8.
2. Check for Unicode replacement characters.
3. Check common mojibake markers using the bundled script list.
4. For Markdown, check that fenced code blocks are balanced.
5. If any check fails, block closeout and repair the document before continuing.

Script

Use the bundled checker when possible:

python .\skills\uth-utf8-guard\scripts\check_utf8_docs.py <path> [<path> ...]

It validates UTF-8 decoding, scans for mojibake markers, and checks Markdown fence parity.

Closeout

If .uth-governance/project.json contains document_language, render the closeout report in that language. For zh-CN, use Chinese headings and Chinese prose; preserve literal paths, commands, skill names, schema values, and code identifiers.

When this guard is used, include:

UTF-8 guard:
- files checked:
- result:
- repaired encoding issues:

If the guard was required but not run, do not claim documentation closeout is complete.

Source: undertaker33/uth-governance — distributed by TomeVault.