What are a token and a context window?
Answer out loud or in your head first, then compare.
Answer
A token is a piece of data in the form the model uses. For text it can correspond to part of a word, a whole word or a character. There's no fixed ratio of “one token is one word”. How text is split depends on the model, the language and the content.
The context window is the limited amount of information the model can take into account in a given call. It holds the instructions, history and tool results, among other things; exactly how the limits are divided depends on the model. That's why pasting a whole repo doesn't always help. It can raise the cost and make it harder to notice the important part. It's better to pass the data you need and leave room for further work.
Why it matters: The basis of managing a coding agent's context and RAG.
Possible follow-ups: Does a bigger window guarantee the model will use every piece of information?
How did it go? Be honest — “Not yet” puts the question on your review list.
Further reading
- Anthropic: Effective context engineering for AI agents(opens in a new tab)29.09.2025. Choosing context, limits, and references instead of blindly growing the input.