How does this skill help manage token costs?

It provides specific strategies for 'concise is key' prompting, including techniques to remove redundancy and move stable content to system prompts to preserve the context window.

What is Chain-of-Thought prompting?

Chain-of-Thought (CoT) is a technique that encourages the LLM to reason step-by-step before providing a final answer, significantly improving accuracy on complex analytical tasks.

Can I use these patterns for sub-agents?

Absolutely. The skill includes a dedicated section on Agent Prompting Best Practices, focusing on setting appropriate degrees of freedom and using persuasion principles.

What is the benefit of Few-Shot learning in this skill?

Few-Shot learning uses 2-5 input-output pairs to show the model the desired behavior rather than just explaining rules, ensuring consistent formatting and edge-case handling.

Prompt Engineering Patterns

Name: Prompt Engineering Patterns
Author: m16khb

bym16khb

データサイエンスとML

Optimizes LLM interactions through advanced prompting techniques like few-shot learning, chain-of-thought, and systematic template design.

概要

This skill provides a comprehensive framework for designing and refining prompts to maximize the performance and reliability of Claude and other LLMs. It offers standardized patterns for few-shot examples, multi-step reasoning traces, and reusable templates, while incorporating Anthropic's best practices for agentic behavior and context window management. Whether you are building sub-agents, automation hooks, or complex Git workflows, this skill ensures high-quality outputs by managing instruction hierarchy, error recovery, and token efficiency.

主な機能

Context window and token efficiency management
Few-shot and Chain-of-Thought reasoning implementations
Production-grade prompt template and variable systems
Persuasion-based agent communication principles
Systematic optimization and A/B testing frameworks
0 GitHub stars

ユースケース

Refining complex RAG-based workflows for better context utilization
Optimizing repetitive automation tasks for consistency and cost-reduction
Designing robust system prompts for specialized sub-agents and hooks

概要

主な機能

Context window and token efficiency management
Few-shot and Chain-of-Thought reasoning implementations
Production-grade prompt template and variable systems
Persuasion-based agent communication principles
Systematic optimization and A/B testing frameworks
0 GitHub stars

ユースケース

Refining complex RAG-based workflows for better context utilization
Optimizing repetitive automation tasks for consistency and cost-reduction
Designing robust system prompts for specialized sub-agents and hooks