prompt-refiner-claude
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseClaude Prompt Refiner
Claude提示词优化指南
Refine prompts to get better results from Claude models by applying Anthropic's recommended patterns.
通过应用Anthropic推荐的模式优化提示词,从而从Claude模型中获得更优结果。
When to Use
适用场景
Invoke this skill when you have a task for Claude that:
- Involves multiple steps or files
- Requires specific output formatting
- Needs careful reasoning or analysis
- Would benefit from structured context
当你有以下类型的Claude任务时,可使用此方法:
- 涉及多个步骤或文件
- 要求特定的输出格式
- 需要严谨的推理或分析
- 能从结构化语境中获益
Refinement Process
优化流程
Step 1: Analyze the Draft Prompt
步骤1:分析草稿提示词
Review the user's prompt for:
- Clear outcome definition
- Sufficient context
- Explicit constraints
- Success criteria
Ask clarifying questions if any of these are missing.
检查用户的提示词是否包含以下要素:
- 清晰的结果定义
- 足够的语境信息
- 明确的约束条件
- 成功标准
如有缺失,提出澄清问题。
Step 2: Apply Claude-Specific Patterns
步骤2:应用Claude专属模式
Structure with XML tags:
Claude responds exceptionally well to XML-style tags for organizing complex prompts:
- - Background information, codebase state, environment
<context> - - The specific action to take
<task> - - Must-have criteria
<requirements> - - Limitations and boundaries
<constraints> - - Sample inputs/outputs if helpful
<examples> - - How to structure the response
<output_format>
Ordering matters:
- Context first (what exists)
- Task second (what to do)
- Requirements third (how to do it)
- Examples last (clarifying edge cases)
使用XML标签结构化:
Claude对用于组织复杂提示词的XML风格标签响应极佳:
- - 背景信息、代码库状态、环境
<context> - - 具体要执行的操作
<task> - - 必须满足的标准
<requirements> - - 限制条件与边界
<constraints> - - 如有帮助可提供输入/输出示例
<examples> - - 响应的结构要求
<output_format>
顺序至关重要:
- 先放语境(现有信息)
- 再放任务(要做什么)
- 接着是要求(怎么做)
- 最后是示例(澄清边缘情况)
Step 3: Enhance for Reasoning
步骤3:增强推理能力
For complex tasks, add thinking prompts:
- "Think through the approach before implementing"
- "Consider these edge cases: ..."
- "Explain your reasoning for key decisions"
对于复杂任务,添加引导推理的提示:
- "先思考实现方法再执行"
- "考虑以下边缘情况:..."
- "解释关键决策的推理过程"
Step 4: Output the Refined Prompt
步骤4:输出优化后的提示词
Present the improved prompt with:
- Clear section headers
- XML tags where beneficial
- Specific, measurable criteria
- An explanation of what changed and why
优化后的提示词需包含以下元素:
- 清晰的章节标题
- 必要时使用XML标签
- 具体、可衡量的标准
- 说明修改内容及原因
Example Transformations
示例转换
Example 1: Vague Task
示例1:模糊任务
Before:
Add caching to the APIAfter:
<context>
The /api/products endpoint currently queries the database on every request.
Average response time is 200ms. We use Redis for other caching in the app.
</context>
<task>
Add Redis caching to the /api/products endpoint to reduce database load.
</task>
<requirements>
- Cache TTL of 5 minutes
- Cache invalidation when products are updated
- Graceful fallback to database if Redis is unavailable
- Add cache hit/miss metrics logging
</requirements>
<constraints>
- Don't change the response format
- Must pass existing integration tests
- Use our existing Redis connection from src/lib/redis.ts
</constraints>优化前:
Add caching to the API优化后:
<context>
/api/products端点目前每次请求都会查询数据库。平均响应时间为200ms。我们在应用中使用Redis处理其他缓存需求。
</context>
<task>
为/api/products端点添加Redis缓存以降低数据库负载。
</task>
<requirements>
- 缓存TTL为5分钟
- 产品更新时自动失效缓存
- Redis不可用时优雅降级到数据库查询
- 添加缓存命中/未命中指标日志
</requirements>
<constraints>
- 不得修改响应格式
- 必须通过现有集成测试
- 使用src/lib/redis.ts中已有的Redis连接
</constraints>Example 2: Code Review Request
示例2:代码审查请求
Before:
Review this PRAfter:
<context>
This PR adds user authentication to our Next.js application.
Our stack: Next.js 14, TypeScript, Prisma, PostgreSQL.
Security is critical - this handles user sessions and passwords.
</context>
<task>
Review the changes in this PR for security issues, code quality, and adherence to our patterns.
</task>
<requirements>
Focus on:
1. Security vulnerabilities (auth bypass, injection, etc.)
2. Error handling and edge cases
3. TypeScript type safety
4. Test coverage for critical paths
</requirements>
<output_format>
Organize your review as:优化前:
Review this PR优化后:
<context>
此PR为我们的Next.js应用添加了用户认证功能。
技术栈:Next.js 14、TypeScript、Prisma、PostgreSQL。
安全性至关重要 - 此模块处理用户会话和密码。
</context>
<task>
审查此PR中的变更,检查是否存在安全问题、代码质量是否达标以及是否符合我们的开发规范。
</task>
<requirements>
重点关注:
1. 安全漏洞(认证绕过、注入攻击等)
2. 错误处理与边缘情况
3. TypeScript类型安全性
4. 关键路径的测试覆盖率
</requirements>
<output_format>
按以下结构组织审查结果:Critical Issues (must fix before merge)
严重问题(合并前必须修复)
Recommendations (should consider)
建议项(应考虑修复)
Minor Suggestions (nice to have)
小建议(可选优化)
What Looks Good (positive feedback)
优秀实践(正面反馈)
</output_format>
undefined</output_format>
undefinedExample 3: Feature Implementation
示例3:功能实现请求
Before:
Add dark modeAfter:
<context>
React application using Tailwind CSS for styling.
Currently only has light mode. Design tokens are in tailwind.config.js.
User preference should persist across sessions.
</context>
<task>
Implement dark mode toggle with system preference detection and persistence.
</task>
<requirements>
- Toggle component in the header
- Detect system preference on first visit
- Persist user choice in localStorage
- Smooth transition between modes
- Update all existing components to support both modes
</requirements>
<constraints>
- Use Tailwind's built-in dark mode support
- Don't add new dependencies
- Ensure WCAG AA contrast ratios in both modes
</constraints>
<examples>
Current light mode colors:
- Background: bg-white
- Text: text-gray-900
- Primary: text-blue-600
Expected dark mode equivalents:
- Background: dark:bg-gray-900
- Text: dark:text-gray-100
- Primary: dark:text-blue-400
</examples>优化前:
Add dark mode优化后:
<context>
使用Tailwind CSS进行样式开发的React应用。
目前仅支持浅色模式。设计令牌存于tailwind.config.js中。
用户偏好应在会话间持久化。
</context>
<task>
实现深色模式切换功能,支持系统偏好检测与持久化存储。
</task>
<requirements>
- 在头部添加切换组件
- 首次访问时检测系统偏好
- 将用户选择持久化到localStorage
- 模式切换时使用平滑过渡
- 更新所有现有组件以支持两种模式
</requirements>
<constraints>
- 使用Tailwind内置的深色模式支持
- 不得添加新依赖
- 确保两种模式均符合WCAG AA对比度标准
</constraints>
<examples>
当前浅色模式颜色:
- 背景:bg-white
- 文本:text-gray-900
- 主色:text-blue-600
预期深色模式对应颜色:
- 背景:dark:bg-gray-900
- 文本:dark:text-gray-100
- 主色:dark:text-blue-400
</examples>Tips for Best Results
最佳实践小贴士
- Be specific about scope - "the auth module" → "src/auth/session.ts"
- Include file paths when relevant
- Reference existing patterns - "follow the pattern in UserService.ts"
- State what NOT to do - constraints prevent unwanted changes
- Define done - what does success look like?
- 明确范围 - "认证模块" → "src/auth/session.ts"
- 相关时包含文件路径
- 参考现有模式 - "遵循UserService.ts中的模式"
- 说明禁止操作 - 约束条件可避免不必要的修改
- 定义完成标准 - 成功的标准是什么?