skill-description-evaluator

Invocation likelihood evaluation assesses how clearly a skill description communicates its purpose to Claude models like Sonnet 4.5 and Haiku. This Skill performs structured sequential-thinking analysis to score semantic clarity, authority level, and user request pattern matching against 0-100 ratings with concrete improvement recommendations.

Yes. This Skill benchmarks your description against competing instructions you provide, revealing alignment gaps and potential conflicts. It outputs comparative 0-100 ratings and actionable guidance to strengthen your description's authority and precision relative to competing prompts.

Effective descriptions balance semantic clarity, explicit authority signaling, and precise user request pattern matching across different Claude model capabilities. This Skill's phase-based workflow identifies gaps in these three areas and delivers specific improvements to maximize invocation reliability on both Sonnet 4.5 and Haiku.

Provide your description text or SKILL.md file path. The Skill returns structured 0-100 ratings paired with concrete recommendations targeting semantic clarity, authority framing, and request pattern specificity. Implement these suggestions to increase invocation likelihood and user intent alignment.

Sonnet 4.5 and Haiku have different reasoning depths and instruction-following patterns. This Skill assesses invocation likelihood separately for each model, revealing where your description succeeds or fails with capability-specific request interpretation, helping you craft descriptions that work reliably across both.

Use this Skill when you need objective, model-specific invocation predictions and concrete improvement guidance. Manual refinement lacks systematic scoring and comparative benchmarking; this automated evaluation reveals blind spots in clarity and authority that affect cross-model reliability.