AI web development · Glossary
Context window
Last updated June 29, 2026 · by Tal Gerafi
A context window is the amount of text — measured in tokens — a large language model can consider at once. Everything the model 'sees' for a task, including instructions, code, and history, must fit inside it.
The context window is the model's working memory for a task. Measured in tokens (chunks of text), it sets a hard limit on how much the model can hold in mind at once — the instructions, the code it's reading, the conversation so far, and the tools' output all share that space.
How it works
When a task's material exceeds the window, something has to give: older content is dropped or summarized, and the model can lose track of earlier detail. This is why what you put in the window matters as much as how big it is. Good practice — sometimes called context engineering — keeps the window focused on the relevant 5% of a codebase rather than stuffing it with everything.
Why it matters for B2B sites
On a real project, the failure mode usually isn't a weak model; it's a model looking at the wrong context. Keeping always-on instructions lean, loading detail on demand, and using subagents with their own windows for noisy work all protect the context budget. That discipline is why an agentic workflow split across focused agents tends to beat one agent trying to hold everything. Greeto's approach is in Building websites with Claude Code.