Inspiration

I wanted to explore a deceptively simple decision. When two production stations break down and only one technician is available, which repair should come first? It’s easy to look at the two repair times. It’s harder to see how that choice will affect every boat still moving through the factory.

For Kiroween, I turned equipment failures into ghosts on the factory floor. When a ghost appears, the affected bay glows and work stops. The player takes over as the production supervisor and decides how to get the line moving again.

What it does

Boat Factory Simulator follows seven boats through seven stations, from hull fabrication to final inspection and launch. The dashboard shows work in progress, completed boats, and how shipments compare with the standard schedule. Players can also open a station’s operating procedure and event history to see what happened.

In the Repair Priority Challenge, Flooring and Decking and Engine and Controls stop at the same time. One technician handles both repairs, and the player chooses the order. In this scenario, Engine first ships all seven boats at 9:15 of simulated factory time. Decking first finishes at 9:25. The two repairs take the same total time in either order, but the shipment schedule changes by ten minutes.

How I built it with Kiro

I gave the Kiro IDE agent a written production specification and worked through the simulator in small steps. For an early Hull Fabrication test, I specified that work should stop after 22 productive minutes, hold at eight minutes remaining during the repair, and finish at minute 40 if repaired immediately. Those expected values gave Kiro a concrete test to implement and gave me a way to check its work.

Kiro then implemented the one-technician repair queue and wrote automated checks for both repair orders. The checks verified the batch finish times, confirmed that repairs never overlapped, and checked that all seven boats shipped. Kiro also compared the standard run against a saved checkpoint to catch regressions.

Challenges and what I learned

The main simulation challenge was keeping productive work, waiting, and repair time separate. A stopped boat has to keep the work it already completed, and a second repair has to wait until the technician is free. If either rule is wrong, the dashboard can show a plausible result even when the schedule is wrong.

The browser test caught a different kind of problem. A Repair now button was still visible after the challenge ended because a style rule overrode its hidden state. Kiro fixed the styling and added a check for that specific bug. Both repair orders were then played through on the public demo in Chrome.

What I took away was the value of giving Kiro measurable outcomes before asking it to change the code. The timing tests showed whether the factory behaved as specified. The browser walkthrough showed whether a player could understand and use it.

Built With

Share this project:

Updates

Submission history