Run an iPhone game in a simulator and log coverage

A cloud agent played an iPhone game in a simulator and returned a recording plus a test report showing what it checked and what it skipped.

Done with Devin· Intermediate

What they did
Cognition had Devin, running on GPT-6 Astra, open Otter Run in an iPhone simulator and exercise the game there. While it worked, the agent captured a recording of the session and then wrote up which checks succeeded, along with the parts it did not cover. Engineers can use those outputs to understand how the game behaves and where manual review is still needed.
How it went
The example did not include pass totals or any timing data. It was presented as an early step toward having engineers inspect less code by hand, though that broader outcome has not been demonstrated yet.
Tips
A report of skipped areas is only partial coverage, so you still need to inspect those parts yourself. Ask for the exact flows you want tested, and tell the agent not to change code.

Try it yourself with Devin

Open [app name] in the simulator and test it as an iPhone game/app. Walk through [key flows to check], and capture a screen recording while you do it. When you finish, give me a report with: checks that passed, checks that failed with repro steps, and anything you did not test. Before making any irreversible action such as submitting, sending, deleting, or purchasing, pause and ask for my approval. Do not modify code.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.