Context Engineering
Rekayasa Konteks
Deciding what goes into the model's context, and what stays out.
Context engineering is the practice of curating what the model sees on each request: which files, which history, which tool definitions, in what order, and how much. It is the successor to prompt engineering, because once a system has tools and memory, the interesting question is no longer the wording of one instruction but the state of the context at the moment of the call.
The core trade is completeness against attention. Everything the model might need costs tokens and competes for attention with the thing you actually want it to do. Good context engineering keeps what is needed, drops what is stale, and makes sure the current instruction is not buried.
A good test: at the moment of the call, could you defend every line of the context? If not, something is there for historical reasons rather than present ones.
Related terms
Context Window
Everything the model sees on one request. Finite, and it decides what fits.
Prompt Engineering
The craft of writing instructions a model follows reliably.
Memory
A system's ability to carry information from one request to the next.
Tool Result
What the harness sends back after running a tool: file contents, command output, an error.
Progressive Disclosure
Show the model only the detail it needs right now, and load more when it asks.
Want this working inside your team?
It starts with one conversation about your workflow.