How does this skill improve token efficiency?

It provides a decision tree to help you select the simplest effective method for your task, preventing over-engineered prompts and unnecessary few-shot examples that consume tokens.

What prompting techniques are included in this skill?

The skill supports over 10 proven methods including Zero-Shot, Few-Shot, Chain of Thought (CoT), ReAct, Tree of Thoughts, and Self-Consistency.

When should I use XML instead of JSON for prompts?

XML is highly recommended for Claude to handle complex hierarchies and instructions, whereas JSON is preferred for API-based interactions and GPT-specific workflows.

Does this work for models other than Claude?

Yes, while it provides specialized XML formatting for Claude, it also includes optimized logic for GPT (JSON), multi-LLM portability, and human-readable YAML formats.

Prompt Engineering Architect

Name: Prompt Engineering Architect
Author: rfxlamia

byrfxlamia

•

Data Science & ML

Designs and optimizes high-performance LLM prompts using advanced reasoning techniques and model-specific structured formats.

The Prompt Engineering Architect skill empowers developers to construct sophisticated AI interactions by intelligently selecting the most effective prompting strategies tailored to task complexity and target models. It automates the decision-making process for choosing between methods like Chain of Thought, ReAct, and Tree of Thoughts, while providing structured formatting guidance—such as XML for Claude or JSON for APIs—to ensure high accuracy, reliability, and token efficiency across diverse LLM workflows.

Key Features

01Intelligent selection of 10+ prompting techniques (CoT, ReAct, ToT)

02Model-optimized formatting (XML for Claude, JSON for GPT/APIs)

03Token efficiency strategies and cost-saving implementation patterns

0442 GitHub stars

05Comprehensive decision matrix based on task complexity and context

06Advanced meta-prompting and multi-step prompt chaining guides

Use Cases

01Refactoring existing prompts to reduce token consumption and improve reasoning

02Transforming complex business logic into structured, high-accuracy LLM prompts

03Designing robust autonomous agent workflows using ReAct and Self-Consistency

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rfxlamia/claude-skillkit prompt-engineering

For use in Claude.ai and ChatGPT

Download Skill