We asked four AI coding agents to rebuild Minesweeper—the results were explosive

aimine2

I can’t tell you how much I appreciate those cording instructions below.

I can’t tell you how much I appreciate those cording instructions below.


Credit: Benj Edwards

execution

This agent not only included the important “cording” feature, but it also included on-screen instructions for using it on both PC and mobile browsers. I “?” Was further impressed by the option of cycling through. Marks when marking intersections with flags, a puzzling feature that seems most human even to me minesweeper Cloners can miss.

On mobile, the option to hold your finger down on a square to mark a flag is a nice touch that makes this the most enjoyable handheld version we’ve tested.

presentation

The old-school emoticon smiley-face button is pretty cute, especially when you explode and get a red “X.” I was less impressed by the playfield “graphics”, which uses a simple “*” for revealed mines and an ugly red “F” for flagged tiles.

The beeps-and-boops sound effects reminded me of my first old-school, pre-sound-blaster PC from the ’80s. This is usually a good thing, but I still appreciate games that give me the option to turn them off.

“Fun” feature

“Surprise: Lucky Sweep Bonus” listed in the corner of the UI explains that clicking the button gives you a free safe tile when available. This can be quite useful in situations where you would otherwise be forced to guess between two tiles that have the same probability of being a mine.

Overall, though, I found it a little strange that the game only gives you this bonus if you find a large, wide area of ​​safe tiles with a single click. It mostly serves as a “Win ​​More” button rather than a feature that provides a good balance of risk vs. reward.

coding experience

OpenAI Codex has a nice terminal interface with features like Cloud Codes (local commands, permission management, and interesting animations showing progress), and it’s quite pleasant to use (OpenAI also offers Codex via a web interface, but we didn’t use it for this evaluation). However, Codex took almost twice as long as Cloud Code to code a functional game, which may contribute to the stronger result here.



<a href

Leave a Comment