Last reviewed
Correct answer: C. A substantial, self-contained piece of content like code or a document
Explanation
Artifacts exist for output you will do something with after the chat is over. Anthropic's test is size and independence: Claude creates one when the content is significant and self-contained — typically more than fifteen lines — and when it is something you are likely to edit, iterate on or reuse outside the conversation. A React component, a migration script, a policy document or a long data table each qualifies, because each still means something once you copy it somewhere else.
Short replies fail that test, and the failure is practical rather than stylistic. A one-line answer, a yes or no, or a clarifying question only makes sense next to the message that prompted it. Lifting one into a panel of its own separates it from the context that made it readable, and costs you a click to read what would otherwise have been visible in the transcript.
The useful question is therefore not whether the answer is important, but whether you will still want it after the chat is closed.
Sources
“It is significant and self-contained, typically over 15 lines.”
Practise 6 questions on this topic
Take Projects & Artifacts — Timed Test (6 questions) — scored instantly, explanation for every question, no login.