Glossary Entry

Context Window

The bounded working memory of an LLM, holding every instruction, file, and tool output the model can currently see. Model performance degrades as it fills, so managing it deliberately is a core skill in agentic work.

LLMs Architecture

Also called: context windows

The context window is the maximum span of tokens a large language model can attend to at once. Everything the model knows about your current task lives there: the system instructions, the conversation so far, the files it has read, and the output of every command it has run. Anything that falls outside is simply invisible to the model.

Two practical consequences follow. First, capacity is finite, so long agent sessions eventually crowd out early instructions with accumulated noise. Second, performance degrades before the hard limit is reached, which is why practitioners clear context between tasks, keep instruction files short, and delegate exploratory reading to subagents with their own separate windows.