Context Window
also context length
The context window is the maximum number of tokens a model can consider at once, including both your input prompt and its generated output.
Analogy
Like a desk that only fits so many papers at a time; anything that doesn't fit has to be left off.
Why it matters
It limits how much data (transcripts, contact records, docs) you can feed into a single Claude call inside an automation before you must summarize or split it.
In practice
A 200K-token context window can hold roughly a 500-page book of text in one request.