muratcankoylan/agent-skills-for-context-engineering — reverse-engineered prompt

I want a polished repo that gives coding agents a reusable set of skills for context engineering. Make it feel like a practical library and plugin pack people can drop into Claude Code, and still understand and reuse on other agent tools too.

The main thing is a collection of clearly named skills that teach agents how to manage limited context well, avoid long session breakdowns, compress or optimize context, design memory and tool use, choose multi agent patterns, and evaluate agent behavior in a production style way. Include practical examples, lightweight Python style pseudocode, and good docs that explain when each skill should activate and what problem it solves.

Please make the experience feel complete, with install and usage guidance, readable documentation, and enough examples that someone building real agent systems can apply it right away. If it fits, also include newer topics like hosted agents and formal belief, desire, intention style reasoning. Keep it platform agnostic in spirit, but make the Claude Code plugin workflow work smoothly. Look up current docs online if you need to.