What is five-dimension-prompt-engine?

This project is an Agent Skill, a kind of add on that gives an AI assistant a specific structured way of working, aimed at everyday people who are not skilled at writing prompts for AI tools. It does two things: turns a vague request into a complete, reusable prompt that can be copied and given to another AI, and evaluates an existing prompt or AI output to find weak spots and suggest concrete improvements. Its core rule is that the prompt itself is the product. By default it only produces a finished prompt, it does not carry out the task described in that prompt. If a user then asks it to actually execute the prompt, the skill steps aside and lets the normal task happen. It is meant for office workers, content creators, social media operators, people using AI image or video generators, and anyone who wants to use AI well but struggles to describe what they need clearly. It is not aimed at people who already know prompt writing tricks. The skill is designed to trigger only in certain situations, such as AI image or video generation prompts, coding or app or website prompts, report and data analysis prompts, research and competitor analysis prompts, business and work plans, and film or content scripts. It intentionally avoids triggering for casual chats, simple factual questions, quick translations, or small edits. In enhancement mode, it checks whether a request fits its trigger conditions, scans the request across five dimensions, then asks the user two or three multiple choice questions to confirm key decisions before generating the final prompt, along with a short explanation of its choices. In evaluation mode, it first lists the weaknesses of an existing prompt or output across those same five dimensions, scores it, shows a diagnostic summary, and proposes an improvement path before producing a revised version once the user agrees. The five dimensions it evaluates are how complete the perspective is, what the user's real underlying need is, how deep the reasoning handles tradeoffs, whether the output is directly actionable, and whether key decisions are explained. The project includes a test set of nine sample scenarios and reports an internal evaluation score of 90.8 out of 100 on its current pre release version.