Smart Prompt

Problem

Ordinary prompts often state only what to do, without goals, constraints, output format, or evaluation criteria. smart-prompt turns a short instruction into a structured prompt and reduces manual copy-paste execution steps.

How It Works

• Intent recognition: uses 5W1H to identify task type, goal, constraints, audience, and output format.
• Scene matching: selects a pattern from code, writing, analysis, creative, data, or general prompts.
• Thinking enhancement: injects commands such as critical evaluation, reverse thinking, first principles, tree reasoning, and constraint optimization.
• Structured generation: outputs a refined prompt with role, task breakdown, few-shot examples, and output format.
• Auto execution: defaults to auto_execute: true, runs the refined prompt immediately, and returns both result and prompt for review.

Boundaries

It is useful for quickly rewriting instructions and trial-running tasks, but not ideal for high-risk operations that require strict human confirmation. Default auto execution reduces confirmation steps, so be cautious with production data, published content, or irreversible actions. The default timeout is 300s, and complex tasks may need splitting. Scoring history and A/B testing compare accuracy, completeness, practicality, and creativity across prompt versions, while final quality still depends on task clarity and model capability.