prompt-engineering

# Prompt Engineering Skill

Expert guidance for crafting effective prompts for LLMs and AI systems using proven techniques and patterns.

## Overview

This skill provides comprehensive expertise for designing, testing, and optimizing prompts across different LLM models and use cases.

## When to Use This Skill

Trigger this skill when:
- Designing prompts for Claude, GPT, or other LLMs
- Optimizing existing prompts for better performance
- Building AI features that require LLM interactions
- Creating system prompts for agents or chatbots
- Implementing chain-of-thought reasoning
- Testing prompt variations for consistency
- Building prompt pipelines or chains
- Designing few-shot learning examples
- Creating role-based AI personas
- Troubleshooting poor LLM outputs

**Keywords:** prompt engineering, LLM, Claude, GPT, few-shot learning, chain-of-thought, prompt optimization, AI prompts, system prompts

## Core Principles

### Prompt Engineering Fundamentals

1. **Clarity and Specificity**: Be explicit about what you want
2. **Context Provision**: Give the model necessary background
3. **Format Specification**: Define exact output format desired
4. **Constraint Setting**: Establish boundaries and guidelines
5. **Example Inclusion**: Show rather than just tell (when appropriate)
6. **Iterative Refinement**: Test and improve based on outputs

### Model Characteristics to Consider

- **Context Window**: How much text the model can process
- **Training Cutoff**: What knowledge the model has
- **Capabilities**: What the model can and cannot do
- **Biases**: Known limitations or tendencies
- **Temperature**: Creativity vs determinism trade-off

## Prompting Techniques

### 1. Zero-Shot Prompting

Direct instruction without examples. Best for simple, well-defined tasks.

```
Analyze the sentiment of this customer review:

Review: "The product arrived quickly and works great. Very satisfied!"

Sentiment:
```

**When to use:**
- Task is straightforward and unambiguous
- Model