prompt-engineer

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Prompt Engineer

Prompt工程师

Overview

概述

Design, test, and optimize prompts for LLM interactions. This skill covers prompt patterns (few-shot, chain-of-thought, ReAct), system prompt design, output formatting, prompt evaluation, and prompt optimization techniques.

为LLM交互设计、测试并优化prompt。本技能涵盖prompt模式（few-shot、chain-of-thought、ReAct）、系统prompt设计、输出格式化、prompt评估以及prompt优化技术。

Features

功能特性

Prompt patterns: few-shot, zero-shot, chain-of-thought, ReAct, self-consistency
System prompt design: role definition, constraints, output format specification
Output formatting: JSON, XML, markdown, structured templates
Prompt evaluation: quality metrics, consistency testing, edge case analysis
Prompt optimization: token reduction, clarity improvement, robustness testing

Prompt模式：few-shot、zero-shot、chain-of-thought、ReAct、self-consistency
系统prompt设计：角色定义、约束条件、输出格式规范
输出格式化：JSON、XML、markdown、结构化模板
Prompt评估：质量指标、一致性测试、边缘案例分析
Prompt优化：token精简、清晰度提升、鲁棒性测试

Usage

使用方法

Identify the user's prompt need (pattern selection, system prompt, output format, or optimization)
Follow the corresponding workflow below
Produce structured outputs: prompt templates, system prompts, output schemas, or evaluation reports

识别用户的prompt需求（模式选择、系统prompt、输出格式或优化）
遵循下方对应的工作流程
生成结构化输出：prompt模板、系统prompt、输出schema、评估报告

Examples

示例

User: "Write a prompt for summarization" Agent: Runs Prompt Design workflow, selects zero-shot pattern, defines role and constraints, produces prompt with output format
User: "Optimize this prompt" Agent: Runs Prompt Optimization workflow, identifies ambiguity, reduces token count, adds clarity, tests edge cases
User: "Evaluate prompt quality" Agent: Runs Prompt Evaluation workflow, tests against quality metrics, identifies failure modes, produces improvement recommendations

用户："Write a prompt for summarization" Agent：执行Prompt设计工作流，选择zero-shot模式，定义角色与约束条件，生成带有输出格式的prompt
用户："Optimize this prompt" Agent：执行Prompt优化工作流，识别模糊点，减少token数量，提升清晰度，测试边缘案例
用户："Evaluate prompt quality" Agent：执行Prompt评估工作流，对照质量指标进行测试，识别失效模式，生成改进建议

When to Use

使用场景

Designing, versioning, and evaluating prompts for LLM-powered features
Building agent workflows (ReAct, tool use, multi-agent coordination)
Optimizing accuracy, format compliance, latency, and token cost
Deploying guardrails, observability, and abuse defenses for GenAI in production

为LLM驱动的功能设计、版本管理并评估prompt
构建Agent工作流（ReAct、工具调用、多Agent协作）
优化准确性、格式合规性、延迟和token成本
在生产环境中为生成式AI部署防护机制、可观测性和滥用防御措施

When NOT to Use

非适用场景

Classical ML model training, feature engineering, or statistical A/B tests → use
```
data-scientist
```
General technical writing, API reference, or runbooks → use
```
tech-writer-researcher
```
Cloud infrastructure, CI/CD, or Kubernetes operations → use
```
infrastructure-engineer
```
Revenue recognition or finance close procedures → use
```
senior-revenue-accountant
```
Multi-feature token reduction roadmap → use
```
ai-token-improvement-plan-engineer
```
Rigorous token-efficiency experiments and ablations → use
```
research-engineer-scientist-tokens
```

传统机器学习模型训练、特征工程或统计A/B测试 → 使用
```
data-scientist
```
通用技术文档编写、API参考或运行手册 → 使用
```
tech-writer-researcher
```
云基础设施、CI/CD或Kubernetes运维 → 使用
```
infrastructure-engineer
```
收入确认或财务结账流程 → 使用
```
senior-revenue-accountant
```
多功能token精简路线图 → 使用
```
ai-token-improvement-plan-engineer
```
严谨的token效率实验与消融研究 → 使用
```
research-engineer-scientist-tokens
```

Core Workflows

核心工作流

1. Prompt Design Workflow

1. Prompt设计工作流

Step-by-step process:

Define the task clearly
- What input does the user provide?
- What output format is required?
- What constraints must be enforced?

Choose the pattern

Pattern	When	Structure
Zero-shot	Simple, well-defined tasks	Instructions + input
Few-shot	Pattern recognition, formatting	Examples + task
Chain-of-thought	Reasoning, math, logic	"Let's think step by step"
Role-based	Domain expertise needed	"You are a senior X..."
Structured	API/programmatic consumption	JSON schema, XML template

Draft and iterate
- Start simple, add complexity only where needed
- Use clear separators (###, XML tags, markdown)
- Specify output format explicitly
- Include constraints and what to avoid
Test with edge cases
- Empty input, malformed input, adversarial input
- Boundary conditions
- Multiple languages or formats

分步流程：

明确任务定义
- 用户提供什么输入？
- 需要什么输出格式？
- 必须遵守哪些约束条件？

选择合适的模式

模式	适用场景	结构
Zero-shot	简单、定义清晰的任务	指令 + 输入
Few-shot	模式识别、格式化任务	示例 + 任务
Chain-of-thought	推理、数学、逻辑任务	"Let's think step by step"
基于角色	需要领域专业知识	"You are a senior X..."
结构化	API/程序化调用	JSON schema、XML模板

草稿与迭代
- 从简单版本开始，仅在必要时增加复杂度
- 使用清晰的分隔符（###、XML标签、markdown）
- 明确指定输出格式
- 包含约束条件及需规避的内容
边缘案例测试
- 空输入、格式错误的输入、对抗性输入
- 边界条件
- 多语言或多格式输入

2. Prompt Optimization & Testing

2. Prompt优化与测试

Evaluation dimensions:

Accuracy: Does it produce correct results? (human or model judge)
Consistency: Same input → same output? (temperature, seed control)
Format compliance: Does output match the schema? (JSON validator)
Latency: Time to first token, total generation time
Cost: Tokens consumed (input + output)

Testing workflow:

Build a benchmark dataset (50-200 diverse examples)
Establish baseline with current prompt
Modify one variable at a time (prompt, model, temperature)
Run A/B comparison on benchmark
Measure and document improvement

评估维度：

准确性：是否生成正确结果？（人工或模型判断）
一致性：相同输入是否产生相同输出？（temperature、seed控制）
格式合规性：输出是否匹配schema？（JSON验证器）
延迟：首token生成时间、总生成时间
成本：消耗的token数量（输入 + 输出）

测试工作流：

构建基准数据集（50-200个多样化示例）
使用当前prompt建立基准线
每次仅修改一个变量（prompt、模型、temperature）
在基准数据集上进行A/B对比
测量并记录改进效果

3. Agent Orchestration

3. Agent编排

Agent patterns:

Pattern	When	Components
ReAct	Tool-using agent	Reasoning + Action + Observation loop
Plan-and-Solve	Multi-step tasks	Planner → Executor → Checker
Reflexion	Self-improvement	Execute → Evaluate → Revise
Multi-agent	Complex workflows	Specialist agents + coordinator

Tool use checklist:

Tool schemas are clearly defined (name, description, parameters)
Agent can handle tool failure gracefully
Tool results are summarized, not passed raw to user
Rate limits and costs are monitored

Agent模式：

模式	适用场景	组件
ReAct	工具调用型Agent	推理 + 动作 + 观察循环
Plan-and-Solve	多步骤任务	规划器 → 执行器 → 检查器
Reflexion	自我提升	执行 → 评估 → 修订
多Agent	复杂工作流	专业Agent + 协调器

工具调用检查清单：

工具schema定义清晰（名称、描述、参数）
Agent可优雅处理工具调用失败
工具结果经过汇总，而非直接传递给用户
监控速率限制与成本