Prompt Optimizer

Problem Solved

Users often say things like “help me build a tool,” “make this website better,” or “understand the code architecture.” These requests usually lack role, goal, context, inputs, and acceptance criteria. If execution starts immediately, an AI or engineer may fill the gaps by guessing and miss the intended outcome. generate-prompt addresses vague inputs by turning a one-line request into a reusable, delivery-ready prompt that tells the next executor what role to assume, what task to complete, under which constraints, and how to judge the result.

How It Works

The skill only generates prompts. It does not read files, explore repositories, call tools, or perform the task. The core workflow is:
- parse the original request to extract the core task and implied intent
- complete missing details using role setup, task description, context information, and instructions/constraints
- choose a general, task-execution, or role-play format based on the scenario
- output a complete prompt that can be pasted and used directly

For complex requests, it can also suggest outlining first, staged execution, checkpoints, visible reasoning, and measurable standards, replacing vague terms such as “fast,” “nice,” or “efficient” with time limits, design standards, complexity bounds, or acceptance criteria. The boundary is clear: when users actually need code changes, file queries, script execution, or deployment, this skill is not an executor. It is best used before work starts to shape requirements into precise instructions, and it can be iterated until the prompt itself is unambiguous.