From Recipe Cards to Full Kitchen

The evolution from prompt engineering to context engineering

2022–2023: Prompt Engineering
📝 RACE
Role
Action
Context
Expectation
📋 RISEN
Role
Instructions
Steps
End goal
Narrowing
🎯 APE
Action
Purpose
Expectation
These checklists taught what information models need—the same principles now applied at system scale.
Same ingredients,
bigger kitchen
2024–2025: Context Engineering
🗄️
Pantry
Knowledge Bases
Long-term storage: docs, embeddings, vector databases
🔪
Prep Station
RAG / Retrieval
Retrieved Docs 📄 Relevant
Vector Search 🔍 Semantic
📖
Recipe Book
Instructions & Features
User-Defined
System Prompt 📖 Custom
Persona / Role 📖 Custom
LLM Features
Reasoning Mode 🔄 Extended
Tool Use ⚙️ Enabled
🔧
Connections
Tools & External Services
Web Search 🌐 Results
Image Creation 🎨 Generated
Code Execution 💻 Output
App Connectors 🔌 MCP/APIs
🍳
Counter
Context Window — What the LLM Sees
Drag items here to build the context window
🧠
Memory
Stored Context
Long-term
User Memory 🧠 Persistent
Short-term
Chat History 💬 Session
User Message 📝 Current
User Query
Orchestrate Context
Generate
iterates until complete
Output
The key insight: The acronyms identified the right ingredients (role, context, examples, constraints). Context engineering is about designing the whole kitchen—systems that assemble the right context automatically for each task.