Last reviewed
Correct answer: A. That the agent judged the work finished — nothing outside the agent weighed the result
Explanation
The principle — a summary is a claim about the work, not a measurement of it.
Why the key is correct — Anthropic states the rule plainly: Claude stops when the work looks done. The glossary gives the consequence: without verification, the only thing deciding the agent is finished is the agent itself. So the summary establishes the agent's own judgement, and nothing more.
Why the others are wrong — looking done triggers the stop, so being done is not a precondition. No check is implied either, which is why the guidance asks Claude to show evidence rather than assert success: the test output, the command it ran and what it returned. And a statement of the outcome is still the agent's own.
Remember this — a completion summary tells you the agent stopped, not that the work is right.
Sources — Anthropic's Claude Code best practices and glossary.
Sources
“Claude stops when the work looks done.”
“without one, the only thing deciding the agent is finished is the agent itself”
“Have Claude show evidence rather than asserting success: the test output, the command it ran and what it returned, or a screenshot of the result.”
Practise 10 questions on this topic
Take Claude Code Workflows — Timed Test 1 (10 questions) — scored instantly, explanation for every question, no login.