Prompt Engineer — Claude Code Skill

A Claude Code custom skill that transforms rough prompts/ideas into production-ready LLM prompts. Supports Claude, GPT, Llama, and other models with advanced techniques like Chain-of-Thought, Constitutional AI, RAG optimization, and more.

Features

• Multi-model support — optimized prompts for Claude, GPT, Llama, Gemini, etc.
• Advanced techniques — CoT, Tree-of-Thoughts, few-shot, Constitutional AI, prompt chaining
• Production-ready output — structured prompts with implementation notes, test cases, and usage guidelines
• Model-specific optimization — leverages each model's strengths (XML for Claude, JSON mode for GPT, etc.)
• Enterprise patterns — templates for agents, RAG, data extraction, and multi-step workflows

Installation

Via npx (recommended)

npx prompt-engineer-skill

Custom install path:

npx prompt-engineer-skill --path ~/.claude/custom-skills/prompt-engineer

Via git

git clone https://github.com/repo-phuocdt/prompt-engineer-skill.git \
  ~/.claude/skills/prompt-engineer

Uninstall

npx prompt-engineer-skill --uninstall

Usage

In Claude Code, invoke the skill:

/prompt-engineer

Then describe what you need:

• "Create a system prompt for a code review agent"
• "Optimize this prompt for RAG retrieval"
• "Design a multi-step reasoning prompt for medical diagnosis"

Output Format

Every response includes:

1. The Prompt — complete, copyable code block
2. Implementation Notes — techniques used, model-specific optimizations, parameter recommendations
3. Testing & Evaluation — test cases, edge cases, optimization suggestions
4. Usage Guidelines — when/how to use, customization options

References

The skill includes curated reference materials:

• references/techniques.md — Advanced prompting techniques catalog
• references/model-specific-optimization-guide.md — Per-model optimization strategies
• references/production-patterns-and-enterprise-templates.md — Enterprise-grade templates

License

MIT — see LICENSE.