Context Engineering
Use this skill when designing or reviewing how context is assembled for AI systems.
Use Cases
- Context stack design
- Retrieval and RAG planning
- Memory design
- Multi-step orchestration
- AI system reliability reviews
Core Principle
Deliver the smallest high-signal context that makes the task solvable and safe.
Layers
Think in layers:
- System policy
- Developer instructions
- Tool contracts
- Runtime state
- Retrieved knowledge
- Memory
- Examples
- Output contract
Repository Mapping
- Root and scoped
AGENTS.md-> developer instructions and runtime context .agents/skills/*-> reusable task workflows and knowledge- Project files and retrieved docs -> knowledge context
- Structured outputs or explicit response formats -> output contract
Companion Resources
context-stack-model.mdmemory-engineering.mdagent-harness-patterns.mdrag-engineering.mdmulti-agent-patterns.mdproduction-checklists.mdreference-templates.mdprivacy-compliance-ai.md