agent-toolkit/skill-judge

Evaluating SKILL.md files for expert value involves scoring knowledge delta, specification compliance, and anti-pattern quality. This skill audits packages to classify content as Expert, Activation, or Redundant, generating a structured report with actionable fixes.

Auditing YAML frontmatter and detecting anti-patterns requires checking structural compliance and pattern recognition. This skill reviews frontmatter quality and flags critical issues, producing a graded report with prioritized recommendations for skill activation readiness.

Scoring specification compliance and usability before publishing involves a multi-dimensional assessment out of 120 points. This skill evaluates progressive disclosure, freedom calibration, and practical usability to generate a final grade and concrete improvement guidance.

Validating workflow usability and knowledge delta for a SKILL.md file categorizes sections to pinpoint what to keep, refine, or remove. This skill estimates the knowledge delta to ensure your package adds genuine expert knowledge rather than redundant content.

The skill evaluation process requires no external dependencies or components. It operates independently to audit SKILL.md files, validating YAML frontmatter and assessing anti-pattern quality to produce a self-contained structured report with scores and grades.

Automated skill evaluation for token efficiency and spec compliance focuses on structural and content analysis rather than runtime behavior. It assesses pattern recognition and progressive disclosure but relies on the provided SKILL.md content to estimate the knowledge delta accurately.