Sandbox demo
Verification: Tests, Browser, Recordings
Agents that prove their work — test runs, live-app screenshots, click-through recordings.
'It should work now' is banned. Antigravity's answer is verification artifacts: the agent runs the tests, boots the app, drives it in Chrome, and attaches the evidence — screenshots and a recording of the flow — to its walkthrough. Your review starts from evidence, not trust.
For UI work, ask for the error paths explicitly: the recording of the happy path plus screenshots of each failure state. Tests catch logic; recordings catch the things tests can't — layout breaks, jank, a spinner that never resolves.
Sandbox transcript
Press Play to replay this transcript step-by-step.
0 / 6
The transcript is canned — no real API calls. The point is to internalize the loop: user → plan → tool → result → answer.
Check your understanding
Q1. Why demand a browser recording for UI work when the tests already pass?
· Score 100% on the quiz.