context-engineering-kit

Why we included this project

Teams that use Claude Code, Cursor, OpenCode, or similar coding agents for real work often find their output varies in quality. This kit collects hand-crafted skills and plugins that add context-engineering techniques directly into the agent's context, such as spec-driven development, subagent-driven workflows, and a reflection loop that reviews finished work and suggests fixes. The plugins are granular, so you install only the pieces you need instead of a heavy all-in-one prompt. They follow the agentskills.io open standard and work across several CLI tools, so they stay portable if your team switches editors. If you want your coding agent to produce more consistent, maintainable code without building your own prompt framework, this is a solid starting point.