Prompt Quality Inspector

What this skill does

• Scoring diagnostics (primary): produces an explainable quality score (0–100) with sub-scores, concrete deductions, risks, and reproducible reasoning.
• Guided optimization (secondary): asks 3–7 prioritized clarification questions, then outputs an execution-ready integrated prompt (requirements + tech + data + security + ops), clearly marking what is confirmed vs assumed.

Design goal: make prompt quality measurable, repeatable, and iteratively improvable.

When to use

• “Score this prompt/spec and tell me what’s missing.”
• “Turn my idea into an industrial-grade prompt that an AI can implement.”
• You have a long requirement dump and worry it is too big / too vague / not executable.
• You want an offline/on-prem, security-heavy system prompt that includes requirements + engineering constraints.

Output contract (pseudo UI)

Always start with one unified "console panel" that uses two logical sections: SCORE and OPTIMIZE.

• In score-only mode, keep OPTIMIZE as a collapsed placeholder (for example: "OPTIMIZE disabled by user mode"), instead of rendering full optimization content.
• In optimize-only mode, render a compact diagnosis summary in SCORE first, then continue in OPTIMIZE.
• In score + optimize mode (default), render both sections fully.