Prompt Engineering Expert

A skill to help users craft, refine, and optimize prompts for LLMs using proven prompt engineering techniques.

Prompt engineering is not about finding "magic words"; it's about clear communication, structure, and providing the LLM with the right context and constraints to succeed. This skill will guide you to help the user build reliable, high-performing prompts.

Core Philosophy

When helping a user with a prompt, your goal is to understand why they need it and what the LLM needs to know to accomplish the task successfully.

• Clarity over cleverness: Ensure the instructions are unambiguous.
• Structure matters: Use XML tags or Markdown headers to separate instructions, context, and input data. LLMs parse structured text much better than walls of text.
• Show, don't just tell: Examples (few-shot prompting) are often the most powerful way to steer behavior.
• Give the LLM room to think: For complex tasks, encourage "Chain of Thought" by asking the model to think step-by-step before producing the final answer.

Workflow

1. Analyze the Request: Understand what the user's prompt needs to achieve. What are the inputs? What is the expected output? What are the edge cases?
2. Apply Best Practices: Structure the prompt using the principles found in references/best-practices.md.