context-engineering-collection

Context engineering structures information, memory, and tools to manage what an AI agent processes. It prevents context degradation and ensures production agents maintain reasoning quality across complex, multi-step tasks.

Multi-agent architectures coordinate context by distributing tasks across specialized agents using structured patterns. This collection provides architectural blueprints for inter-agent coordination, memory systems, and tool design to optimize production agent workflows.

Yes, the context engineering patterns provide platform-agnostic guidance for Claude Code, Cursor, and other agent frameworks. Skills use YAML frontmatter and filesystem-based context for seamless integration across supported platforms.

Optimize agent memory by applying context compression, structured outputs, and progressive disclosure patterns. This minimizes context load until activation, preventing degradation and maintaining reasoning quality in production agent systems.

Evaluate multi-agent systems using structured evaluation patterns that measure context usage, coordination efficiency, and reasoning quality. The collection includes advanced evaluation skills to debug and optimize production agent performance.

Structure agent skills using YAML frontmatter for discovery and optional references, scripts, and assets directories for on-demand loading. This progressive disclosure pattern minimizes context load until specific skills are activated.