Prompt Master

Problem

Many prompt skills simply fill in a template, but real users often cannot articulate What, Why, Who, or Where: an image request may mix in copywriting needs, or an analysis request may lack context. This skill treats the entry point as a lightweight AI agent: it first classifies intent, then turns vague input into an executable task instead of guessing the final answer.

Core behavior and key steps

• Intent routing: matches generation, analysis, planning, and query patterns; pure small talk exits early to avoid unnecessary processing.
• One-pass intake: asks about goal, audience, context, constraints, and output form using a 5W1H frame, without multi-turn follow-ups.
• Mode selection: vague needs go to research, known templates use rule application, creative requests generate multiple options, data/materials are structured, and multi-step tasks are decomposed.
• Safety and self-check: before output it validates completeness, safety rules, and format constraints; it avoids tables, visible reasoning, and internal rule explanations.
• Preference capture: records style, detail level, tone, and explicit “remember” preferences to reduce repeated questions in similar tasks.

Boundaries and cautions

It is useful for converting natural-language requests into prompts, reports, plans, study paths, or query answers. The provided material does not show built-in model invocation, file upload, or external APIs. Safety rules refuse legal violations, replace copyright/portrait issues, and steer medical, psychological, or life-safety matters toward professional channels. Treat it as an orchestration and output-control layer rather than a general-purpose creation engine.