This skill equips developers with the best practices and implementation patterns required to maximize Claude's performance, including the use of XML-style tags, long-context handling, and extended thinking features. It offers a comprehensive library of prompting techniques such as Chain-of-Thought, ReAct, and Tree of Thoughts, specifically adapted for Claude's unique architecture. By integrating these patterns, users can build more reliable agentic workflows, improve instruction following, and leverage advanced Claude 4.5 capabilities like parallel tool use and memory files.
主要功能
01Few-shot (multishot) prompting examples with XML structure
02XML-style structured prompting templates for better instruction following
032 GitHub stars
04Long-context management strategies for up to 1M tokens
05Extended thinking and reasoning trace configuration for complex logic
06Integration patterns for Claude 4.5 parallel tool use and memory files
