OpenParable
Skip to content
← Back to skills

Taiwanmd Validate

Testing & Quality
taiwanmd-validate

Taiwanmd Validate is an Agent Skill for testing & quality published in the frank890417/taiwan-md repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Validate Taiwan.md articles for editorial quality, frontmatter correctness, and content standards. Use when reviewing PRs that touch knowledge/ or src/content/, writing new articles, or when the user asks to check article quality. Runs frontmatter validation, word count check, and reports issues. TRIGGER when: user says "validate", "check article", "review quality", or when editing .md files in knowledge/ or src/content/.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/frank890417/taiwan-md/blob/HEAD/.claude/skills/taiwanmd-validate/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/taiwanmd-validate, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

taiwanmd-validate: Article Quality Validation

Validate Taiwan.md articles against editorial standards, frontmatter schema, and content quality requirements.

Validation steps

1. Frontmatter validation

cd "$(git rev-parse --show-toplevel)" && node scripts/test-frontmatter.mjs

Checks all articles for:

  • Required fields: title, description, date, tags
  • Valid date format
  • Tags is an array
  • No duplicate slugs
  • Valid file naming

2. Check a specific article

Read the article and verify manually:

# Read the article
cat "knowledge/<Category>/<filename>.md" | head -30

Check:

  • title — present and descriptive
  • description — 1-2 sentences, contains the "counter-intuitive core insight"
  • date — valid ISO date
  • tags — array with 3-8 relevant tags
  • featured — boolean (optional)
  • lastHumanReview — set to true if human-reviewed
  • lastVerified — date of last fact-check

3. Content quality check (article-health prose-health)

cd "$(git rev-parse --show-toplevel)" && python3 scripts/tools/article-health.py "knowledge/<Category>/<filename>.md" --check=prose-health

Scores articles 0-10 for AI-generated "hollow" content. Score > 4 = needs rewrite.

Checks for:

  • Bullet list padding (lazy structure)
  • Missing specific dates/years
  • Hollow modifiers ("非常", "極其", "值得注意的是")
  • Excessive em-dashes (> 4 per article = AI signal)
  • Textbook-style openings
  • Missing source references

4. Reference check

cd "$(git rev-parse --show-toplevel)" && node scripts/check-references.mjs

Verifies articles have "參考資料" or "延伸閱讀" sections with URLs.

5. Wikilink check

cd "$(git rev-parse --show-toplevel)" && node scripts/test-wikilinks.mjs

Validates all [[wikilink]] cross-references resolve to existing articles.

Quality standards (EDITORIAL v4)

Every article should:

  1. Open with a named person — not an institution or abstract concept
  2. Have a counter-intuitive core insight in the description
  3. Include specific dates and numbers — not vague references
  4. Cite sources — 參考資料 section with URLs
  5. Word count ≥ 1,500 for full articles
  6. Use wikilinks [[Article Name]] for cross-references
  7. Multi-perspective for sensitive topics (politics, identity)

When to use

  • After writing or editing an article
  • When reviewing a PR that modifies knowledge/ files
  • When the user asks "is this article good enough?"
  • Before marking lastHumanReview: true