# Cinderbelt uses a bot to test its economy and a person to decide if it is fun

> Finding a trade loop that prints money and judging whether trading feels worthwhile call for different tests.

Published 2026-11-08

Canonical: https://machines-on-machines.anth.us/posts/cinderbelt-test-bot-and-human-playtesting/

::image{src="assets/cinderbelt-test-bot-and-human-playtesting/cinderbelt-game.png" alt="A player ship just outside the ring-shaped Origin Base station in the browser game Cinderbelt" credit="Image: Cinderbelt contributors, CC BY 4.0 for game art, via Cinderbelt."}

A player ship leaving Origin Base in Cinderbelt 0.3.4. This gameplay capture shows the simulation described in the post. [Image source](https://github.com/amoreo140/cinderbelt/blob/main/LICENSE-ASSETS.md). [Licence](https://creativecommons.org/licenses/by/4.0/).

Cinderbelt's development notes describe two kinds of playtest. A scripted pilot flies the game through many reproducible starting conditions and checks details such as how quickly money accumulates. A separate mode tries to exploit the rules. Then the person directing the project plays it to judge whether the result feels fair, readable and fun.

The creator says coding agents implement changes under human direction and that large changes get automated checks and separate review. A reproducible simulation helps because a broken situation can be replayed without flying the whole journey again.

The public repository lets readers inspect the notes and tooling. Its division of work is especially clear on one point. A test can show that a route generates unlimited money. Deciding whether the route makes a good game still depends on what the creator wants players to experience.

[Read the source](https://github.com/amoreo140/cinderbelt/blob/main/docs/MADE-WITH-AI.md).
<!-- papyrus-source-version: 3 -->