/context: predict, then check

Build-along · ~8 min at your own keyboard

You met /context in the terminal tour. This time, no notes and no hints: predict your own numbers before you look, then explain where you were wrong. Being able to predict what fills the window is what lets you change it on purpose — that's the point of this whole track.

The long version: Understanding /context ↗

Do this

0/4

Checkpoints

▸You should now see your written prediction before you run anything.
▸You should now see your prediction next to the real /context numbers.
▸You should now see an explanation for each big miss.

The three ways this goes wrong

Numbers look huge before you type. — That is normal system and tool overhead. Note the biggest three lines; that is the point of the lesson.
/context output is cut off or hard to read. — Widen the terminal window, or ask Claude to restate the numbers in a table.
You compare a fresh session to a different folder. — Use the same folder for both runs so only the messages change.

Where these ideas come from: Token, Context window (AI Hero dictionary). Paraphrased here.

Share what happened

Paste your written predictions, the actual /context numbers from both runs, and one line per big miss explaining it.

Strip API keys, passwords, and tokens before you paste. Swap them for [KEY].

🔒 Don't paste confidential, personal-health, or payment info.