Asset Validation

Systematic validation of Claude Code AI assets. Use after batch creation or modification and before accepting assets into the shared package baseline.

When to Use

• After creating or modifying one or more Claude assets
• During a package-wide validation pass
• When investigating broken references or activation issues

Validation Layers

1. File Size

Every hot-path Markdown asset should stay within Claude-friendly size limits.

2. Frontmatter

Check required fields:

• rule files use the frontmatter expected by the Claude runtime
• SKILL.md contains name and description
• CLAUDE.md stays plain Markdown

3. Cross-References

Verify:

• @skill-name targets an existing folder in skills/
• agent references target an existing file in agents/
• hook script paths resolve under hooks/scripts/
• settings.json commands reference valid scripts

4. Runtime Readiness

Confirm:

• assets assume the .claude/ runtime layout
• hook configs use .claude/hooks/scripts/...
• settings.json remains consistent with hook scripts
• no installer-specific commands remain

5. Naming

• agent files use kebab-case.md
• skill folders use kebab-case/
• template files use *.claude.template.md

Report Format

Return:

• Summary
• Passed checks
• Findings
• Required follow-up

Use validation-checklist.md for the detailed manual checklist.