prompt-engineer

SKILL.md

This skill enables comprehensive prompt engineering across multiple LLM models. Engineer, optimize, and refine prompts using established best practices. Create new prompts from scratch or improve existing ones for maximum effectiveness. Recommend optimal models based on specific requirements through interactive analysis.

| XML Tags | 3+ components, structured output | High - clear separation | | Role Prompting | Domain expertise needed | Medium - contextual knowledge | | Clear & Direct | Always (baseline) | Critical - foundation | | Multishot Prompting | Format/style consistency | High - 40-60% improvement | | Chain of Thought | Complex reasoning | High - accuracy boost |

| Prompt Chaining | Multi-stage workflows | High - manages complexity |

當用戶要求“創建提示”、“優化提示”、“改進此提示”、“設計提示”、“提示工程最佳實踐”、“使此提示更好”、“推薦模型”、“我應該使用哪個模型”、“最佳模型”、“GPT 與 Claude”、“Opus 與 Sonnet”、“Haiku 與 Sonnet”、“分析提示質量”、“修復我的提示”時，應使用此技能。提示”、“提示 Claude”、“提示 GPT”，或者需要任何 LLM 的提示工程技術、模型選擇或提示優化方面的幫助（Claude Opus/Sonnet/Haiku 4.5、GPT 5.1/Codex、Gemini Pro 3.0）。 來源：ibutters/claudecodeplugins。

查看原文