Prompting

What it does

This skill assists users in crafting effective prompts for large language models like Claude, GPT, Gemini, and Llama. It provides a structured workflow and templates to guide prompt creation, focusing on clarity, specificity, and desired output formats. The goal is to maximize the quality and relevance of responses from these AI models by optimizing the input prompts.

When to use it

• When you need to generate summaries or analyses from text content.
• For code review tasks requiring specific security, performance, or maintainability checks.
• When you want to enforce a particular output format like Markdown, JSON, or XML.
• To improve the consistency and accuracy of responses from AI models.
• When dealing with complex reasoning or multi-step workflows that benefit from techniques like chain-of-thought prompting.

Key capabilities

• Prompt Templates: Provides structured templates for prompt creation including optional context/role, required task definition, constraints, output format, and examples.
• Technique Guidance: Offers a quick reference to prompting techniques such as few-shot learning, chain-of-thought reasoning, role prompting, task decomposition, and constraint specification.
• Output Format Specification: Supports defining desired output formats including XML, JSON, and Markdown.
• Troubleshooting Guide: Includes a guide for diagnosing common prompt issues (vague outputs, wrong format, missing details, inconsistent results, off-topic responses) with suggested solutions.
• Improvement Checklist: Provides a checklist to evaluate the effectiveness of existing prompts.

Example prompts

• "Summarize this article in 3 bullet points." (Minimal Prompt)
• "You are a senior code reviewer. Review this code for: Security vulnerabilities, Performance issues, Maintainability concerns. Format your response as: ## Summary [1-2 sentences] ## Issues - [severity]: [description] ## Recommendations [prioritized list]" (Structured Prompt)

Tips & gotchas

• Specificity is key: Replace vague instructions like "be concise" with concrete limits like "under 100 words."
• Format matters: Explicitly define the desired output format to ensure consistency and usability.
• Explore techniques: Refer to references/techniques.md for detailed patterns and examples of prompting techniques.