prompt
**Prompt Engineering Fundamentals**
**What is Prompt Engineering?**
Prompt engineering is the practice of crafting effective inputs to large language models to guide them toward desired outputs. It is both an art and a science that significantly impacts LLM performance.
**Core Prompting Techniques**
**Zero-Shot Prompting**
Directly state what you want without examples:
```
Summarize the following article in 3 bullet points:
[article text]
```
**Few-Shot Prompting**
Provide examples to guide the output format:
```
Translate English to French:
- Hello → Bonjour
- Goodbye → Au revoir
- Thank you → Merci
- How are you? →
```
**Chain-of-Thought (CoT)**
Encourage step-by-step reasoning:
```
Solve this math problem step by step:
If a train travels 120 miles in 2 hours, what is its average speed?
```
**ReAct (Reasoning + Acting)**
Combine reasoning with tool use:
```
Question: What is the population of Tokyo?
Thought: I need to search for current Tokyo population data.
Action: search["Tokyo population 2024"]
Observation: Tokyo metropolitan area has 37.4 million people.
Answer: The population of Tokyo metropolitan area is approximately 37.4 million.
```
**Prompt Structure Best Practices**
1. **Be specific**: "Write a 300-word professional email" not "Write an email"
2. **Use delimiters**: XML tags or markdown to separate sections
3. **Specify format**: JSON, bullet points, or structured output
4. **Set persona**: "You are an expert software architect..."
5. **Include examples**: Show desired input-output pairs
**Common Mistakes**
- Vague instructions leading to inconsistent outputs
- Not specifying output format
- Missing context or constraints
- Over-complicated prompts that confuse the model