/context anatomy

Tens of thousands of tokens are spent before your first message: system prompt, tool definitions, memory, rules. Then a listing's full data, a stack of comps, and a page of client notes pile on top. /context itemizes it. You can't manage what you've never looked at, and most people never look. Everything here is counted in tokens: the chunks the model reads and is billed by, roughly three quarters of a word each in English. Window size, cost and speed are all measured in them.