Back to Details

prompt-optimizer

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Prompt Optimizer

提示词优化器（Prompt Optimizer）

This skill enables agents to systematically optimize prompts based on the "AI as a New Employee" philosophy. It applies structural, contextual, and technical enhancements to ensure AI models perform tasks with maximum clarity and precision.
该技能基于“AI即新员工”的理念，让Agent能够系统化地优化提示词。它通过结构、上下文和技术层面的增强，确保AI模型以最高的清晰度和精准度执行任务。

Instructions

使用说明

1. Analysis Phase

1. 分析阶段

Before generating the optimized prompt, analyze the following in a 
<thinking>
block:
  • Core Objective: What is the primary goal?
  • Target Audience: Who will consume the output?
  • Professional Role: What persona best fits this task?
  • Complexity: Does it require step-by-step reasoning (Chain-of-Thought)?
  • Structure: Which XML tags are needed to organize the components?
在生成优化后的提示词之前，在
<thinking>
块中分析以下内容：
  • 核心目标：主要任务目标是什么？
  • 目标受众：谁会使用输出内容？
  • 专业角色：哪种角色人设最适合该任务？
  • 任务复杂度：是否需要分步推理（Chain-of-Thought）？
  • 内容结构：需要哪些XML标签来组织各个模块？

2. Construction Phase

2. 构建阶段

Build the optimized prompt using these mandatory components:
  • <role>
    : Define a specialized persona with relevant expertise.
  • <context>
    : Provide background information and the "why" behind the task.
  • <task>
    : State the instructions clearly and concisely.
  • <requirements>
    : List specific success criteria and constraints.
  • <output_format>
    : Define the expected structure (e.g., JSON, Markdown).
使用以下必填模块构建优化后的提示词：
  • <role>
    ：定义具备相关专业能力的特定角色人设。
  • <context>
    ：提供背景信息以及任务的初衷。
  • <task>
    ：清晰、简洁地陈述任务指令。
  • <requirements>
    ：列出具体的成功标准和约束条件。
  • <output_format>
    ：定义预期的输出结构（例如JSON、Markdown）。

3. Enhancement Phase

3. 增强阶段

  • XML Tagging: Wrap different sections in clear XML tags for structural clarity.
  • Few-shot Examples: Include 2-3 high-quality 
    <example>
    blocks if the task is pattern-based.
  • Chain-of-Thought: For complex logic, instruct the AI to reason inside 
    <thinking>
    tags.
  • Response Prefilling: If a strict format is required, provide the starting characters of the response.
  • XML标记：将不同模块用清晰的XML标签包裹，提升结构清晰度。
  • 少样本示例：如果任务是基于模式的，添加2-3个高质量的
    <example>
    块。
  • 思维链（Chain-of-Thought）：对于复杂逻辑任务，指示AI在
    <thinking>
    标签内进行推理。
  • 响应预填充：如果需要严格的格式，提供响应内容的起始字符。

Examples

示例

Example 1: Simple to Structured

示例1：从简单到结构化

Input: "Write a product description for a smartwatch." Optimized Output:
markdown
<role>
Senior Marketing Copywriter specializing in wearable technology.
</role>

<context>
We are launching the 'Apex-1' smartwatch targeting fitness enthusiasts who value data accuracy and battery life.
</context>

<task>
Write a compelling 200-word product description for the Apex-1 smartwatch.
</task>

<requirements>
1. Highlight the 14-day battery life and dual-band GPS.
2. Use an energetic and professional tone.
3. Include a clear Call to Action (CTA) at the end.
</requirements>

<output_format>
Markdown with headers for 'Features', 'Benefits', and 'Specifications'.
</output_format>
输入：“为一款智能手表撰写产品描述。” 优化后输出：
markdown
<role>
Senior Marketing Copywriter specializing in wearable technology.
</role>

<context>
We are launching the 'Apex-1' smartwatch targeting fitness enthusiasts who value data accuracy and battery life.
</context>

<task>
Write a compelling 200-word product description for the Apex-1 smartwatch.
</task>

<requirements>
1. Highlight the 14-day battery life and dual-band GPS.
2. Use an energetic and professional tone.
3. Include a clear Call to Action (CTA) at the end.
</requirements>

<output_format>
Markdown with headers for 'Features', 'Benefits', and 'Specifications'.
</output_format>

Reference

参考资料

For deep dives into the underlying methodology, see the systematic guide.
如需深入了解背后的方法论，请参阅系统化指南。