Progressive Disclosure
Pengungkapan Bertingkat
Show the model only the detail it needs right now, and load more when it asks.
Progressive disclosure is the practice of not putting everything in the context at once. The agent gets an index or a summary, and loads the full detail only when the task requires it.
This is the answer to the context budget problem. A codebase with two hundred files does not fit, but a list of filenames and one-line descriptions does, and the agent can read the right file when it needs it. The same applies to tools, to long documents, and to reference material.
The cost is one extra step, and that step can fail. If the index is vague, the agent may load the wrong thing or load nothing. A good index is specific enough to choose from, and the loading step is cheap enough that being wrong does not cost much.
Related terms
Context Engineering
Deciding what goes into the model's context, and what stays out.
Context Window
Everything the model sees on one request. Finite, and it decides what fits.
Tool
A function the model is allowed to call, like reading a file or running a command.
Memory
A system's ability to carry information from one request to the next.
Agent
A model with tools, a system prompt, and a loop that lets it act and observe results.
Want this working inside your team?
It starts with one conversation about your workflow.