skill_evaluator

Run scripts/validate_skill.py on the skill directory for automated structural checks, then manually score six dimensions (Naming, Description, Content Quality, Structure, Degrees of Freedom, Anti-Patterns) using the weighted rubric to produce an evaluation report.

A good description states what the skill does and when to use it, written in third person under 1024 characters. It must include specific activation triggers like 'Use when working with PDF documents' rather than vague phrases like 'A skill for PDFs.'

Skill names must be lowercase letters, numbers, and hyphens only, maximum 64 characters, with no reserved words like 'anthropic' or 'claude' and no XML tags. Gerund form such as 'processing-pdfs' is preferred over vague names like 'pdf-tool'.

The final score is a weighted average: Content Quality 30%, Structure 25%, Description 20%, Naming 10%, Degrees of Freedom 10%, and Anti-Patterns 5%. Scores of 4.5-5.0 indicate the skill is ready for publication.

Common anti-patterns include presenting too many options without a recommendation, time-sensitive date conditionals, inconsistent terminology, Windows-style backslash paths, deeply nested references, scripts that punt error handling, and unjustified magic numbers.

The SKILL.md body should stay under 500 lines, file references must be only one level deep, and reference files longer than 100 lines need a table of contents. All paths must use forward slashes.