What is context engineering?
Context engineering is the practice of designing and managing everything an AI model sees when it works, including instructions, retrieved documents, conversation history, memory, tool definitions and user permissions, so it has the right information to perform a task reliably.
Context engineering vs prompt engineering
| Prompt engineering | Context engineering | |
|---|---|---|
| Focus | Wording of instructions | All information given to the model |
| Scope | Single prompt | Whole system |
| Includes | Instructions, examples | Retrieval, memory, tools, state |
Techniques
Retrieve only relevant data, summarize long histories, structure tool outputs, and filter by permissions.
Why it matters for agents
Agents fail more often from missing or noisy context than from model limits.
