The problem

Vague, one-liner instructions lead to unpredictable AI outputs that require constant back-and-forth. When critical context like role, audience, and constraints is missing, LLMs default to generic and low-value responses.

What it does

• Refactors colloquial requests into structured, professional instructions using a Role, Task, Context, Steps, Format, and Constraints framework.
• Fills context gaps by making reasonable, industry-standard assumptions instead of stalling with follow-up questions.
• Applies internal chain-of-thought diagnostics to determine the user's true underlying goal and necessary expert persona.
• Encodes anti-hallucination boundaries and measurable goals directly into the generated prompt.

Frameworks & tools

Optimized for LLMs like Claude 3.5 Sonnet, GPT-4o, and specialized coding or writing agents.

Why this beats prompting it yourself

Writing a comprehensive prompt from scratch is time-consuming and often misses technical constraints like tone calibration or specific output formatting. This skill automates the architecture process, ensuring every instruction you give an AI is high-fidelity and execution-ready in seconds.

Use cases

• Converting a vague "write a workout plan" into a detailed fitness coach prompt with specific physiological constraints.
• Transforming "help with a LinkedIn post" into a high-conversion social media strategist prompt.
• Refining raw business ideas into structured pitch deck outlines for executive-level personas.
• Turning loose coding requests into precise technical specifications for AI software engineers.

Known limitations

This skill generates the prompt itself rather than executing the underlying task. It requires the user to provide at least a basic intent or topic to function effectively.