Context Engineering Principles

This skill provides a comprehensive framework for managing the finite attention budget of Large Language Models to prevent accuracy degradation. It focuses on maximizing signal density while minimizing context rot by offering specific patterns for tool design, agent architecture, and multi-agent coordination. Whether you are building Model Context Protocol (MCP) servers or complex autonomous systems, this skill ensures that Claude maintains high recall and precision through advanced techniques like progressive disclosure, context compaction, and structured external memory.

주요 기능

사용 사례