Agent-Skills-for-Context-Engineering
A comprehensive collection of Agent Skills for context engineering, multi-agent architectures, and production agent systems. Use when building, optimizing, or debugging agent systems that require effective context management.
A collection of 14 installable Claude Code skills that teach and implement context engineering patterns, helping AI agents manage what information fits in their limited attention window for better outcomes.
This repository is a collection of Agent Skills focused on a practice called context engineering — the discipline of managing what information gets fed into an AI language model's limited attention window. Unlike prompt engineering (which focuses on writing good instructions), context engineering deals with the full picture: system prompts, tool definitions, retrieved documents, conversation history, and tool outputs. The core insight is that as context grows longer, models start to struggle and miss information buried in the middle, so good context engineering means finding the smallest set of high-quality tokens that produce the best outcomes.
The collection is organized into roughly four categories. Foundational skills cover concepts like context degradation and compression. Architectural skills address how to build multi-agent systems, design memory storage (short-term, long-term, and graph-based), and use filesystems for storing agent state. Operational skills cover topics like evaluation frameworks and LLM-as-a-Judge techniques for measuring quality. A newer cognitive architecture skill covers formal BDI (Beliefs-Desires-Intentions) modeling for deliberative AI agents.
The skills are designed to work with Claude Code as a plugin marketplace — you can install all 14 skills in a single command and they activate automatically based on your task. The patterns are also platform-agnostic and intended to transfer across tools like Cursor and other agent platforms. The full README is longer than what was provided.
Where it fits
- Install all 14 skills with a single command so Claude Code automatically applies context compression and memory patterns based on your current task.
- Learn multi-agent memory architecture patterns including short-term working memory, long-term persistent storage, and graph-based memory design.
- Apply LLM-as-a-Judge evaluation frameworks to measure and improve the quality of your AI agent's outputs.